Application of siRNA-loaded hydrogel in combination with radiotherapy and chemotherapy in the treatment of GBM

Abstract

The application belongs to the technical field of tumor treatment, and discloses application of siRNA-loaded hydrogel in cooperation with radiotherapy and chemotherapy in treatment of GBM. The application develops a triple-responsive intracavity hydrogel (HPAC-siNTRK2@Gel), which can realize precise silencing of TrkB in residual tumor cells in the resection cavity through cRGD-mediated tumor targeting, redox-triggered siRNA release and pH-sensitive degradation. In a syngeneic mouse model and a patient-derived xenograft model, the therapeutic hydrogel significantly inhibits the progression of glioma after surgical resection and prolongs the survival period. More importantly, it synergizes with radiotherapy to inhibit tumor invasion at the brain-tumor interface, further prolonging the survival period. In summary, this work establishes a precise treatment strategy based on neural circuit information for the treatment of glioblastoma.

Description

Technical Field

[0001] This invention belongs to the field of tumor treatment technology, and more specifically, relates to the application of siRNA-loaded hydrogels in synergistic radiotherapy and chemotherapy in the treatment of GBM. Background Technology

[0002] Glioblastoma (GBM) is the most aggressive primary brain tumor in adults. Despite advancements in treatments such as surgery, radiotherapy, and chemotherapy, its mortality rate remains high. The current standard of care involves maximal resection while ensuring functional safety, followed by concurrent chemoradiotherapy based on temozolomide (TMZ). However, the median overall survival for GBM is less than 18 months, and almost all patients eventually experience local recurrence. A key reason for this treatment predicament lies in the unique anatomical and physiological limitations of brain tissue: the blood-brain barrier, relatively limited immune surveillance, and the unavoidable trade-off between "tumor resection extent" and "functional area preservation," all of which collectively limit the clinical efficacy of traditional treatments.

[0003] Glioma cells do not exist in isolation; they can form functional excitatory synaptic connections with neurons and respond to neuronal activity through calcium-permeable AMPA receptors and neurotrophic factor signaling. Neuronal-derived molecular factors can promote tumor proliferation, invasion, and neural circuit remodeling; conversely, tumor-induced neural circuit dysfunction can lead to cognitive impairment and tumor-associated epilepsy. In addition to these functional consequences, synaptic activity can also induce the formation of neuronal-like membranous protrusions, known as "tumor microtubes," which interconnect glioma cells into an invasive network and participate in the formation of radiotherapy and chemotherapy resistance. Therefore, there is an urgent need for an intervention strategy that can selectively disrupt the pathological tumor-neuronal malignant synaptic interaction while preserving physiological synaptic function as much as possible. Summary of the Invention

[0004] Based on the aforementioned deficiencies in existing technologies, this invention first discovers through research that tumor-neuronal synapses may constitute potential therapeutic targets. Based on this, we propose a postoperative strategy: combining surgical resection with local treatment to specifically inhibit the establishment of synaptic connections between residual tumor cells and neurons at the GBM-brain interface.

[0005] Based on this, the present invention combines it with radiotherapy to inhibit tumor invasion at the brain-tumor interface and further prolong survival.

[0006] The objective of this invention is achieved through the following technical solution:

[0007] Small interfering RNA (siRNA) offers a promising strategy for targeted tumor therapy due to its potent gene silencing capabilities. To efficiently and selectively deliver TrkB-targeting siRNA (siNTRK2) to residual glioma cells at the tumor-brain interface, we selected hyperbranched poly(amidoamine) (HPAA) as the core nanocarrier material. HPAA possesses a high density of amine groups, facilitating the formation of complexes with siRNA; the introduction of disulfide bonds into its polymer backbone allows for reductive degradation in the glutathione (GSH)-rich reducing environment within tumor cells. Unlike traditional high-molecular-weight cationic polymers or lipid formulations, which may lead to prolonged retention and nonspecific membrane toxicity, HPAA rapidly cleaves into low-molecular-weight, amine-containing fragments after intracellular endocytosis, thereby reducing the adverse effects of continuous cation exposure. This design makes HPAA more suitable for "transient and localized" gene silencing interventions in the postoperative brain microenvironment.

[0008] We first synthesized the HPAA polymer backbone via a Michael addition reaction of N,N′-bis(acryloyl)cystamine and aminoethylpiperazine. Then, we used the abundant amino groups on HPAA to perform an amidation reaction with Mal-amido-PEG1-C2-NHS to obtain HPAA-PEG-MAL. Further, we performed a Michael addition reaction between the maleimide group of HPAA-PEG-MAL and the thiol-containing HS-PEG-CRGD to obtain HPAA-PEG-cRGD (HPAC). The cRGD group can achieve tumor targeting by recognizing integrin αvβ3 on the surface of GBM cells. Finally, we complexed siNTRK2 with HPAC to form the HPAC-siNTRK2 nanocomposite.

[0009] One of the main challenges of intraoperative cavity therapy is that locally administered formulations can be rapidly cleared by cerebrospinal fluid (CSF). Given the good biocompatibility, injectability, adaptability to irregular cavities, and ability to achieve sustained and stimulus-responsive release, hydrogels have become a promising platform for local drug delivery.

[0010] Therefore, the present invention first provides the application of siRNA-loaded hydrogels in the preparation of drugs for the treatment and / or adjuvant treatment of GBM, wherein the hydrogel is obtained by mixing OHA, CMC and HPAC-siNTRK2 in a certain proportion, and the formulation of the hydrogel is 3% OHA + 5% CMC + 3% HPAC-siNTRK2.

[0011] The preparation process of HPAC-siNTRK2 is as follows:

[0012] S1. Synthesis of HPAA-PEG-MAL: HPAA undergoes an amidation reaction with Mal-amido-PEG1-C2-NHS to obtain HPAA-PEG-MAL;

[0013] S2. Synthesis of HPAC: HPAA-PEG-MAL and HS-PEG-CRGD undergo a Michell addition reaction to yield HPAC;

[0014] S3, HPAC-siNTRK2 complex: The HPAC-siNTRK2 complex is obtained by mixing and incubating the HPAC and siNTRK2 described in S2; wherein the mass ratio of HPAC to siNTRK2 is 20~80:1;

[0015] The sequence of siNTRK2 described in S3 is shown in SEQ ID NO: 1 and SEQ ID NO: 2.

[0016] Preferably, in the above applications, the siRNA-loaded hydrogel is used in combination with / or radiotherapy and / or TMZ.

[0017] Preferably, in the above applications, the GBM is mesenchymal type G0 147 or classical type G0 927.

[0018] Preferably, in the above applications, the combination of the siRNA-loaded hydrogel and / or radiotherapy, and / or TMZ, compared to any single treatment, has at least one of the following effects:

[0019] (1) Prolong the survival of GBM after surgery, or;

[0020] (2) Delaying postoperative GBM tumor recurrence, or;

[0021] (3) Promotes postoperative apoptosis of GBM tumor cells, or;

[0022] (4) Reduce postoperative GBM tumor cell proliferation, or;

[0023] (5) Reduce tumor invasion caused by radiotherapy, or;

[0024] (6) Disrupts the functional synaptic coupling between tumor cells and neurons after surgery.

[0025] More preferably, the promotion of postoperative GBM tumor cell apoptosis refers to the promotion of cleaved caspase-3 in tumor cells. + Apoptosis.

[0026] More preferably, the reduction of postoperative GBM tumor cell proliferation refers to the reduction of Ki67 in tumor cells. + proliferation.

[0027] More preferably, the disruption of the functional synaptic coupling between tumor cells and neurons after surgery refers to reducing the fluorescence intensity of Synapsin-1 and PSD95 at the tumor-brain interface.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] This invention develops a triple-responsive intracavitary hydrogel (HPAC-siNTRK2@Gel) that achieves precise silencing of TrkB in residual tumor cells within the surgical cavity through cRGD-mediated tumor targeting, redox-triggered siRNA release, and pH-sensitive degradation. In both syngeneic mouse models and patient-derived xenograft models, this therapeutic hydrogel significantly inhibited glioma progression and prolonged survival after surgical resection. More importantly, it synergistically interacts with radiotherapy to inhibit tumor invasion at the brain-tumor interface, further extending survival. In summary, this work establishes a precision treatment strategy based on neural circuit information for the treatment of glioblastoma. Attached Figure Description

[0030] Figure 1The tumor-neuronal synaptic communication at the GBM-brain interface suggests potential targets for local therapy; A–C, representative paraffin-embedded brain tissue sections from en-bloc specimens of glioblastoma patients, showing the necrotic core, tumor-brain interface, distal cortex, and distal white matter; the tissue regions were delineated by experienced neuropathologists based on the histopathological features of hematoxylin-eosin (H&E) stained sections (B); A, multiplex immunofluorescence staining of the above regions: Nestin (tumor cells, red), NeuN (neuronal cell bodies, green), PSD95 (postsynaptic marker, white), and Synapsin-1 (presynaptic marker, magenta); B, corresponding H&E stained sections, showing the anatomical boundaries used for region annotation; C, high-power view of the tumor-brain interface, necrotic core, distal cortex, and distal white matter, corresponding to the boxed area in A; scale bar: 2 mm (A, B) and 20 mm. μm (C); D, Quantitative analysis of Nestin, NeuN, PSD95, and Synapsin-1 fluorescence intensity in each region; E–F, Quantitative analysis of presumed excitatory synapses based on the colocalization of Synapsin-1 and PSD95 puncta signals; E, Number of colocalized synaptic puncta structures; F, Mean size of colocalized puncta structures; All quantitative analyses in A–F were based on a sample of 6 independent glioblastoma patients (n = 6); Data are expressed as mean ± standard deviation (sd); Statistical differences were analyzed using one-way ANOVA and Tukey post-hoc test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.001). 0.0001); G, dot plot of single-cell RNA sequencing (scRNA-seq) analysis, showing the transcriptional characteristics of tumor cells in different GBM regions, highlighting NTRK2 expression and genes related to synaptic structure and signaling, calcium channels and glutamate receptors; TC, tumor core; PTB, brain tissue surrounding the tumor;

[0031] Figure 2 High-magnification images of different glioblastoma-related regions are shown; the high-magnification images show the necrotic core, the tumor-brain junction, the distant cerebral cortex, and the distant white matter region, which correspond to the regions marked in Figure 1A. These regions were determined based on the hematoxylin-eosin staining results in Figure 1B; multiplex immunofluorescence staining with Nestin (red), NeuN (green), PSD95 (white), and Synapsin-1 (purple) was performed; scale bar: 50 micrometers;

[0032] Figure 3A-3D represents a quantitative analysis of the distribution of synaptic markers in different glioblastoma-related regions; A, number of Synapsin-1 dots in each region; B, mean size of Synapsin-1 dots; C, number of PSD95 dots in each region; D, mean size of PSD95 dots. One-way ANOVA combined with Tukey's post-hoc test was used for statistical comparison (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Data are presented as mean ± standard deviation and were derived from a sample of 6 independent glioblastoma patients (n = 6). Figure 3E-3H is a detailed anatomical depiction of the tumor parenchyma, invasive margins, and distant cortex in a glioblastoma specimen; E. A low-magnification overview of a representative paraffin-embedded glioblastoma section obtained through en bloc resection; FH, representing representative regions corresponding to the distant cortex (F, green dashed box), invasive margins (G, blue dashed box), and tumor parenchyma (H, red dashed box), as shown in Figure E; region boundaries were determined based on histopathological features in hematoxylin-eosin (H&E) stained sections; for each region (FH), a representative H&E image (left) and a multiplex immunofluorescence image (right) are shown; multiplex staining was performed against Nestin (tumor cells, red), NeuN (neuronal cell bodies, green), PSD95 (postsynaptic marker, white), and Synapsin-1 (presynaptic marker, purple); scale bar: 2 mm (E) and 200 mm. Micrometer (FH); IK, Multiplex immunofluorescence analysis of glioblastoma tumor at the junction with the cerebral cortex; I, Multiplex immunofluorescence staining of the same region, including Nestin (red), NeuN (green), PSD95 (white), and Synapsin-1 (purple); J, High-magnification image showing the colocalization of Nestin (red) and NeuN (green); K, High-magnification image showing the colocalization of Synapsin-1 (presynaptic label, purple) and PSD95 (postsynaptic label, white) in different regions; Scale bar: 2 mm; LQ, CGGA-based programmed transcriptomic analysis of synaptic-related genes; L heatmap showing the expression patterns of genes related to synaptic structural signaling, calcium channels, and glutamate receptors in TrkB high-expression and low-expression groups; Color scale represents association effect, ranging from -1 (blue) to 1 (red); (M - Q) The violin plot shows the ssGSEA enrichment scores of calcium channels, glutamate receptors, and synaptic signaling pathways in the TrkB (M), NLGN3 (N), TSP1 (O), CHRM3 (P), and AMPAR (Q) pathways. Samples are divided into high expression group (red) and low expression group (blue) according to median expression level; statistical significance was determined by Wilcoxon rank-sum test; NS indicates no significant difference; when a significant difference is observed, the P value will be marked;

[0033] Figure 4This paper presents the design, synthesis, and characterization of HPAC-based siRNA nanocarriers. A, Schematic diagram of the synthesis of HPAC_siNTRK2 and its GSH-responsive disassembly. The preparation process of HPAC is as follows: first, HPAA is synthesized, then sequentially coupled with Mal-amido-PEG1-C2-NHS and HS-PEG-cRGD, and further complexed with siNTRK2 to form the HPAC_siNTRK2 nanocomposite. B, ^1H NMR spectra of HPAA, HPAA-PEG, and HPAC. C, Representative transmission electron microscopy (TEM) images of HPAC nanoparticles. D, Hydration size (DLS) analysis and high-magnification TEM image of HPAC, scale bar: 50 nm. E, Measurement of Zeta potential of HPAC and HPAC-siRNA complex before and after incubation with 10 mM GSH for 30 min (n = 3). F, Gel retardation under different HPAC:siRNA mass ratios. G, Gel permeation chromatography (GPC) curves of HPAC with and without GSH incubation, showing changes in peak time; H, Cytotoxicity of HPAA, HPAC, and PEI in GL261 cells (n = 5); I, Flow cytometry quantitative results of HPAA / FAM-siRNA and HPAC / FAM-siRNA transfection efficiency under different mass ratios (n = 3); J, Flow cytometry analysis of transfection efficiency of FAM-siRNA delivered by different vectors in GL261 cells under serum-containing conditions; K, Quantitative transfection efficiency corresponding to J (n = 3); L, Confocal laser scanning microscopy (CLSM) images of GL261 cells treated with FAM-siRNA delivered by different vectors under the same conditions as J, scale bar: 50 μm; M, Representative Western blots. Blot images showing the expression of TrkB, pTrkB, ERK, and pERK in GL261 cells after transfection with PEI / siNTRK2 or HPAC / siNTRK2; quantitative analysis of protein expression levels corresponding to N and M (n = 3); O, schematic diagram of the separation of primary mouse astrocytes and patient-derived GBM cells; P, CLSM images and magnified views of patient-derived primary GBM cells (Human_GBM_926) and newborn mouse primary astrocytes after HPAC-Cy5 / siRNA_FAM treatment, scale bars: 20 μm (top) and 5 μm (bottom); data in E, H, I, K, and N are expressed as mean ± standard deviation (sd). Statistical differences were analyzed using one-way or two-way ANOVA, and Tukey's multiple comparison test was performed (*P < 0.05, ****P < 0.0001).

[0034] Figure 5 The comparison evaluation of HPAC and LNP delivery of siRNA is shown below; A, flow cytometry analysis of GL261 glioma cells after incubation with Cy5-labeled HPAC nanocarriers for 0.5, 2, 4, and 8 h; B, quantitative proportion of Cy5-positive cells corresponding to A (n = 3); C, transmission electron microscopy (TEM) images of SM-102 lipid nanoparticles (LNPs), scale bar: 100 nm; D, dynamic light scattering (DLS) analysis of LNP hydration particle size; E, LNP zeta potential measured in PBS (pH 7.4); F, cell viability of GL261 cells after treatment with HPAC, HPAA, or LNPs at increasing material concentrations for 24 h (CCK-8 assay, n = 5); G, confocal laser scanning microscopy (CLSM) images of GL261 cells treated with FAM-labeled siRNA delivered by LNPs, HPAA, or HPAC under serum-containing conditions, scale bar: 50 nm. μm; H, representative flow cytometry histogram showing FAM-siRNA fluorescence signal in GL261 cells after different vector treatments; I, quantification of siRNA transfection efficiency corresponding to H (n = 3); J, representative Western blot image showing TrkB protein expression in GL261 cells after LNPs, HPAA, or HPAC-mediated delivery of siNTRK2; K, grayscale quantification of TrkB protein level, normalized with GAPDH (n = 3); L, flow cytometry analysis of Cy5 fluorescence intensity in GL261 cells after treatment with HPAC_Cy5, HPAC_Cy5 / siRNA, or HPAA_Cy5, respectively; M, flow cytometry analysis of Cy5 fluorescence intensity in primary neonatal mouse astrocytes after treatment with the same formulation as L; N, quantification of average Cy5 fluorescence intensity corresponding to L and M (n = 3) O, CLSM images of GL261 cells treated with HPAC_Cy5 / FAM-siRNA, showing the distribution of Cy5-labeled HPAC and FAM-labeled siRNA within the cells. Scale bars: 20 μm (top) and 5 μm (bottom). P, High-magnification CLSM images of GL261 cells, primary human GBM cells (Human_GBM_926), and primary mouse astrocytes treated with HPAC_Cy5 / FAM-siRNA, with linear fluorescence intensity curves along the indicated line segments to demonstrate the spatial distribution of Cy5 and FAM signals within single cells. Data in B, E, F, I, K, and N are expressed as mean ± standard deviation (sd). Statistical differences were analyzed using one-way or two-way ANOVA, and Tukey's multiple comparison test was performed (**P < 0.01, ***P < 0.001, ****P < 0.0001).

[0035] Figure 6 The design and characterization of a pH-responsive hydrogel for tumor microenvironment-adapted siRNA delivery are shown below: A, Schematic diagram of the synthesis of a pH-responsive hydrogel encapsulating the HPAC-siNTRK2 complex; B, Gelation process of HPAC-siNTRK2@Gel; C, Injectability demonstration of HPAC-siNTRK2@Gel using a 0.45 mm inner diameter syringe; D, Shape reshaping and adaptability of HPAC-siNTRK2@Gel under different mold shapes; E, Mechanical characterization of five different hydrogel formulations: E, Changes in storage modulus (G′) and loss modulus (G″) over time under 1% strain; F, Amplitude scan: Response of G′ and G″ as shear strain increases from 0.1% to 1000%; G, Curve of hydrogel shear viscosity as a function of shear rate, demonstrating shear thinning properties; H, Macroscopic self-healing properties of HPAC-siNTRK2@Gel: Demonstrating dye diffusion across the damaged interface, 30 I. Tensile test of the composite gel after min, and schematic diagram of dynamic Schiff base bonds promoting self-repair; I. ​​Visual observation of the composite process after the gel was sheared into small fragments, combined with cyclic strain test (1%–500%); J. Swelling and degradation behavior of HPAC-siNTRK2@Gel in PBS at different pH (6.0, 6.5, 6.8 and 7.4) (n = 3 biologically independent experiments); K. Scanning electron microscopy (SEM) images of HPAC-siNTRK2@Gel after degradation in PBS at pH 6.0 and pH 7.4; L. Adhesion ability of HPAC-siNTRK2@Gel to various mouse organs (brain, heart, liver, spleen, lung, kidney); M. Representative immunofluorescence images of neuroinflammatory response in the perioperative brain region on day 7 after surgery in the GL261 model, including untreated control, surgery-only control, and control treated with OHA / CMC+HPAC after surgery. Gel-treated group, brain slice staining: GFAP (green), IBA-1 (red), and DAPI (blue); N–O. Quantitative analysis of the proportion of IBA-1 and GFAP positive areas in the peri-resectable region of the above groups was performed on postoperative days 3, 7, and 28. Data are expressed as mean ± standard deviation (sd) (n = 5). Statistical differences were analyzed using two-way ANOVA and Tukey's multiple comparison test (*P < 0.05, ***P < 0.001, ****P < 0.0001).

[0036] Figure 7The image shows an assessment of peri-resectable nerve inflammation and neurotoxicity following OHA / CMC+HPAC hydrogel implantation; A, representative immunohistochemical staining images of IL-1β in the peri-resectable region on days 3, 7, and 28 post-treatment in the GL261 model, comparing the control group, the surgery-only group, and the OHA / CMC+HPAC gel-treated group; B, quantitative proportion of IL-1β-positive area corresponding to A (n=3); C, representative immunohistochemical staining images of TNF-α in the peri-resectable region on days 3, 7, and 28 post-treatment in the indicated groups; D, quantitative proportion of TNF-α-positive area corresponding to C (n=3). 3) E, Representative Nissl staining images of pericaval cortical neurons in the control group, the surgery-only group, and the OHA / CMC+HPACGel treatment group on days 3 and 28 post-treatment; F, Representative Nissl staining images of hippocampal neurons in the indicated groups on day 14 post-treatment; G, Representative immunofluorescence images of NeuN (green) and DAPI (blue) in the pericaval cortex on days 3, 7, and 28 post-treatment; H, Quantitative density of NeuN-positive cells corresponding to G (n = 5); Data are expressed as mean ± standard deviation (sd). Statistical differences were analyzed using two-way ANOVA and Tukey's multiple comparison test (*P < 0.05).

[0037] Figure 8A shows the systemic toxicity evaluation of the intraoperative cavity hydrogel formulation after treatment; A, representative H&E stained tissue sections of major organs (heart, liver, spleen, lung, kidney) 15 days after intraoperative cavity administration of different formulations (PBS, HPAC_siNTRK2, Gel, HPAC_Negsi@Gel, HPAA_siNTRK2@Gel, HPAC_siNTRK2@Gel) in GL261 orthotopic glioma-bearing mice, scale bar: 100 μm; B, blood biochemical analysis of liver and kidney function indicators of the same animal, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), albumin (ALB), total bilirubin (TBIL), blood urea nitrogen (BUN), and creatinine (CREA). Data are expressed as mean ± standard deviation (n=3). Statistical comparisons were performed using one-way ANOVA and Tukey's multiple comparison test; Figure CG shows that the drug-free OHA / CMC hydrogel does not promote glioma cell migration and proliferation. C, E: Representative images (C) and quantitative results (E) of GL261 cell migration after 12 and 24 h of incubation in control culture medium or OHA+CMC hydrogel extract, using Transwell migration assay (crystal violet staining), scale bar: 50 μm (C), n=5; D, F: Representative images (D) and quantitative results (F) of GL261 cell migration after 0, 12, and 24 h of incubation in control culture medium or OHA+CMC hydrogel extract, scale bar: 100 μm (D), n=5; G: Viability of GL261 cells at different time points after incubation in control culture medium or OHA+CMC hydrogel extract using CCK-8 assay, n=5; Data in E, F, and G are expressed as mean ± standard deviation. Statistical comparisons were performed using two-way ANOVA and Tukey multiple comparison test.

[0038] Figure 9The experiment demonstrates the long-term in vivo retention of hydrogel within the surgical cavity and its responsive degradation by the tumor. A: Schematic diagram of the experimental design for long-term in vivo tracking of hydrogel retention after tumor resection. GL261 tumors expressing luciferase were orthotopically seeded and surgically removed, followed by a single intraoperative injection of hydrogel loaded with FAM-labeled siRNA. In vitro fluorescence imaging of brain tissue was performed on postoperative days 3, 10, 20, and 30. B: Representative in vitro fluorescence images of brain tissue at the indicated time points for different treatment groups defined in A, showing residual FAM-siRNA signals within the resection cavity. C: Quantitative analysis of the fluorescence intensity of in vitro FAM-siRNA within the resection cavity corresponding to B on postoperative days 3, 10, 20, and 30 (n = 3). D: Representative confocal laser scanning microscopy (CLSM) images of brain tissue sections collected on postoperative day 30; cell nuclei were counterstained with DAPI (blue), scale bar: 50 μm. Data in C are expressed as mean ± standard deviation (sd), and statistical differences were analyzed using one-way ANOVA. ANOVA was performed and Tukey's multiple comparison test was conducted (*P < 0.05, **P < 0.01).

[0039] Figure 10The results show that intraoperative HPAC-siNTRK2@Gel inhibits tumor recurrence and prolongs survival in an orthotopic glioma model; A, Flowchart of the experimental procedure for establishing the orthotopic Luci⁺-GL261 GBM mouse model; On day 8 after inoculation, tumor-bearing mice were randomly divided into 6 treatment groups, and T1-weighted FLAIR MRI scans were used to locate the tumor and guide surgical resection; B, Representative intraoperative photographs showing tumor resection and the application of HPAC-siNTRK2@Gel hydrogel. C. Green boxes with arrows indicate resected tumor tissue; D. H&E-stained brain slices on postoperative day 15 showing tumor recurrence (n = 3 biologically independent experiments); E. Kaplan–Meier survival curves (n = 10 per group); F. MRI images at different time points before and after surgery, with yellow boxes indicating the preoperative frontal lobe GBM tumor localization area to guide surgical resection; green boxes delineating postoperative residual resection cavities; red boxes indicating postoperative recurrent tumors; F. Quantitative analysis of tumor recurrence volume based on MRI measurements (n = 3 per group); G. TUNEL immunofluorescence image of GFP-GL261 tumor on postoperative day 15; H. Quantification of TUNEL positive signal using Mander's colocalization coefficient (n = 3); I–J. Ki67 staining and quantitative analysis of Ki67⁺ cell count (n = 3); K. Orthotopic patient-derived glioma stem cell (GSC) xenograft model (G0 Representative bioluminescence imaging (BLI) at specified time points before and after HPAC-siNTRK2@Gel administration in the surgical cavity for mice carrying G0147 (mesenchymal GBM) and GSC2 (anterior neural GBM); L, Kaplan-Meier survival analysis of mice carrying G0147 or GSC2 xenografts after "surgery + HPAC-siNTRK2@Gel" or corresponding control treatment; M, Quantitative analysis of BLI signal intensity before and after treatment in GSC xenograft models (n = 3); Data in D and L were analyzed using the log-rank (Mantel-Cox) test; other comparisons were performed using one-way or two-way ANOVA and Tukey post-hoc test. All values ​​are expressed as mean ± standard deviation (sd) (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

[0040] Figure 11A) Shows the sustained intraoperative antitumor efficacy and tolerability of HPAC-siNTRK2@Gel in an orthotopic glioma model; A) Representative longitudinal bioluminescence imaging (BLI) images of orthotopic GL261-Luc glioma-bearing mice after surgical resection and intraoperative treatment with the indicated formulation, at the indicated postoperative time points; B) Quantitative analysis of BLI signal intensity corresponding to A (n = 3 per group); C) Weight monitoring of GL261 tumor-bearing mice after surgery and treatment (n = 6 per group); D) Representative BLI images of orthotopic patient-derived glioma stem cell (GSC) xenograft models (G0 927, classic GBM; GSC 1203, mixed GBM) at specified time points before treatment and after surgery (with or without intraoperative administration of HPAC-siNTRK2@Gel); E) Mice carrying G0 927 or GSC 1203 xenograft tumors at the time of surgery + Kaplan-Meier survival analysis of HPAC-siNTRK2@Gel or the corresponding surgical control group (sample size is shown in the figure); quantitative analysis of BLI signal intensity at specified time points before and after treatment in xenograft models F, G0 927 and GSC 1203 (n = 3); data in E were analyzed using the Mantel-Cox test; other comparisons were performed using two-way ANOVA and Tukey's multiple comparison test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001), all data are expressed as mean ± standard deviation (sd); Figure 11 G represents the transcriptomic subtyping of patient-derived glioma stem cell lines. The heatmap shows the gene expression scores of three classic glioma molecular subtypes (classical, mesenchymal, and proneuronal) from batch RNA sequencing of four patient-derived glioma stem cell lines (G0927, G0147, GSC1203, and GSC2). Rows represent subtype characteristics, columns represent samples, and colors represent standardized scores (Z-scores), with red indicating high scores and blue indicating low scores. The top color marker indicates the molecular subtype (classical, mesenchymal, proneuronal, or mixed) assigned to each sample. If the difference between the highest and second-highest scores is >0.5, the sample is assigned to the corresponding highest-scoring subtype; otherwise, it is marked as "mixed".

[0041] Figure 12The following are functional and structural assessments of the neuron-glioma synaptic interface in the GL261 model: A. Schematic diagram of the electrophysiological recording paradigm; GFP-labeled GL261 glioma cells were stereotactically seeded into the CA1 region of the mouse hippocampus, bipolar electrodes were placed in the Schaffer collateral pathway to stimulate excitatory afferent fibers, and whole-cell patch-clamp recordings of GFP⁺ GL261 cells were performed in acute hippocampal sections; B. Representative microscopic images of GFP⁺ GL261 glioma cells clamped during whole-cell patch-clamp recording, with biocytin contained in the recording electrodes; C. Representative confocal images of GL261 cells perfused with biocytin during whole-cell recording; Biocytin (red) and GFP-expressing GL261 cells (green) are co-localized, and the cell nuclei are counterstained with DAPI (blue); D. Representative voltage-clamp recordings of GFP⁺ GL261 cells during paired electrical stimulation of Schaffer collateral fibers (stimulation interval 1 s), with a cell holding potential of −70°. mV, stimulation artifacts before synaptic response have been removed; E–F, quantitative analysis of the amplitude (E) and half-width (F) of EPSC induced by the first and second stimuli in D, each point represents a recorded GL261 cell, data from a total of 6 tumor-bearing mice; G, immunogold electron microscopy image showing the ultrastructural contact between neurons and GL261 glioma cells in vivo, gold particles labeled with GFP to identify GL261 cells (white arrow), presynaptic neuronal terminals containing synaptic vesicles (blue arrow) adjacent to the GL261 cell membrane, synaptic cleft and postsynaptic density visible (yellow arrow); H, immunofluorescence staining of phosphorylated TrkB (p-TrkB, red) at the tumor-brain interface in the GL261 model, left: low magnification image showing the tumor boundary (dashed line); right: high magnification image of the tumor-adjacent area and non-tumor area, cell nuclei counterstained with DAPI (blue); statistical analysis of E and F was performed using paired two-sided Student's t test (*P < 0.05). Data are expressed as mean ± standard deviation (sd);

[0042] Figure 13The diagram shows the effects of intraoperative HPAC-siNTRK2@Gel treatment on disrupted functional neuron-glioma synaptic interactions; A, a schematic diagram of the patch-seq experimental procedure for evaluating tumor-neuron synaptic interactions in a GFP⁺ GL261 glioblastoma mouse model, where RNA was extracted from the same cell after whole-cell patch-clamp recording and biocytin perfusion, and then used for single-cell RNA sequencing (scRNA-seq) analysis; B, a representative microscopic image of relapsed GFP⁺ GL261 glioblastoma cells clamped by whole-cell recording electrodes in an acute brain slice; C, a confocal image (red, Alexa 555) of GFP⁺ GL261 glioblastoma cells perfused with biocytin after patch-seq recording; D, (top left) a schematic diagram of the effect of treatment on tumor-neuron synaptic interactions; (top right) a representative single-cell transcriptome profile of a single GFP⁺ GL261 glioblastoma cell; (bottom) GFP⁺ under different treatment conditions. Representative voltage-clamp recordings of spontaneous excitatory postsynaptic currents (sEPSCs) in GL261 glioblastoma cells; representative sEPSC trajectories corresponding to the regions marked by the red dashed boxes in E and D; F, a dot plot obtained based on patch-seq analysis, highlighting GFP⁺ under different treatment conditions. Transcriptional differences in GL261 glioblastoma cells (PBS control: 20 cells from 6 mice; HPAC-siNTRK2@Gel treatment: 19 cells from 6 mice); G–H, quantitative analysis of tumor-neuronal synapse-related electrophysiological parameters, including G, EPSC amplitude; H, EPSC frequency (PBS control: 36 cells from 6 mice; HPAC-siNTRK2@Gel treatment: 35 cells from 6 mice); I–J, scatter plot showing the correlation between sEPSC amplitude (I) and frequency (J) of glioma cells and a “synapse-related gene expression score” constructed based on genes in F, reflecting the association between synaptic activity and transcriptomic features (PBS control: 20 cells from 6 mice; HPAC-siNTRK2@Gel treatment: 19 cells from 6 mice); K, patient-derived G0 Representative voltage-clamp recordings of sEPSCs in 147 (mesenchymal GBM) xenografts under PBS control or HPAC-siNTRK2@Gel treatment conditions; the segments shown above each trajectory at higher temporal resolution are marked by red dashed boxes; Quantitative analysis of sEPSC frequency and amplitude in L,G0 147 xenografts (PBS control: 30 cells from 7 mice; HPAC-siNTRK2@Gel: 32 cells from 8 mice);M, Representative voltage-clamp recordings of sEPSCs in patient-derived GSC2 (preneural GBM) xenografts under PBS control or HPAC-siNTRK2@Gel treatment conditions; fragments shown at higher temporal resolution are indicated by red dashed boxes; N, Quantitative analysis of sEPSC frequency and amplitude in GSC2 xenografts (PBS control: 33 cells from 8 mice; HPAC-siNTRK2@Gel: 28 cells from 7 mice); O–P, Representative multiplex immunofluorescence images of the tumor-brain interface of G0 147 (O) and GSC2 (P) xenografts on postoperative day 28, showing Nestin (glioblastoma cells, red), NeuN (neurons, green), PSD95 (postsynaptic punctate structures, yellow), and Synapsin-1 (presynaptic punctate structures, magenta), with nuclei counterstained with DAPI (blue); Scale bar: 100 μm; Data in G, H, L, and N are expressed as mean ± Standard deviation (sd) is expressed as a whole, and statistical analysis was performed using a two-tailed unpaired Student's t-test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

[0043] Figure 14Electrophysiological and multiplex immunofluorescence analyses were shown in other glioma models following intraoperative HPAC-siNTRK2@Gel treatment; A, representative voltage-clamp recordings of spontaneous excitatory postsynaptic currents (sEPSCs) in patient-derived G0 927 (classic GBM) xenografts under PBS control or intraoperative HPAC-siNTRK2@Gel treatment conditions, with red dashed boxes indicating segments displayed at higher temporal resolution above each trajectory; B, G0 Quantitative analysis of sEPSC frequency and amplitude in 927 (classical GBM) xenografts under PBS control and HPAC-siNTRK2@Gel treatment conditions (PBS control: 27 cells from 6 mice; HPAC-siNTRK2@Gel: 25 cells from 6 mice); C, Representative voltage-clamped records of sEPSCs in patient-derived GSC1203 (mixed subtype GBM) xenografts under PBS control or intraoperative HPAC-siNTRK2@Gel treatment conditions, with red dashed boxes indicating fragments displayed at higher temporal resolution; D, Quantitative analysis of sEPSC frequency and amplitude in GSC1203 (mixed subtype GBM) xenografts under PBS control and HPAC-siNTRK2@Gel treatment conditions. Quantitative analysis of sEPSC frequency and amplitude under siNTRK2@Gel treatment (PBS control: 35 cells from 7 mice; HPAC-siNTRK2@Gel: 30 cells from 6 mice); E, Representative multiplex immunofluorescence images of the tumor-brain interface adjacent to the surgical cavity on postoperative day 15 in the GL261 model, comparing the PBS control and the HPAC-siNTRK2@Gel treatment group within the surgical cavity; Staining markers included Nestin (red), NeuN (green), PSD95 (white), and Synapsin-1 (magenta); cell nuclei were counterstained with DAPI (blue); the boundary of the surgical cavity was delineated with dashed lines; F–G, G0 Representative multiplex immunofluorescence images of the tumor-brain interface on day 28 post-operatively for 927 (classic GBM) xenograft tumors: PBS control group (F) and intraoperative HPAC-siNTRK2@Gel treatment group (G). Staining markers included Nestin (red), NeuN (green), PSD95 (yellow), and Synapsin-1 (magenta). Cell nuclei were counterstained with DAPI (blue). Data in B and D are expressed as mean ± standard deviation (sd). Statistical comparisons were performed using a two-sided unpaired Student's t-test (*P < 0.05).

[0044] Figure 15The top image shows the synthesis of HPAC-PEG-cRGD (HPAC): HPAC is obtained by reacting hyperbranched poly(amidoamine) (HPAA) with Mal-amido-PEG1-C2-NHS and HS-PEG-CRGD; subsequently, it is compounded with siNTRK2 and mixed with oxidized hyaluronic acid (OHA) and carboxymethyl chitosan (CMC) to form the HPAC-siNTRK2@Gel hydrogel system; the bottom image shows the mechanism of action of postoperative intraoperative injection of HPAC-siNTRK2@Gel: 1) Tumor recurrence 1) The hydrogel disassembles and liquefies in response to the acidic tumor microenvironment, releasing the encapsulated HPAC-siNTRK2 nanodelivery system; 2) HPAC-siNTRK2 targets the tumor by recognizing integrin αvβ3 on the surface of GBM cells through cRGD ligands; 3) After the nanocomposite enters GBM cells, the high intracellular concentration of glutathione (GSH) triggers the release of siRNA from HPAC-siNTRK2; 4) The released siNTRK2 downregulates the expression of TrkB receptors in tumor cells, blocks the BDNF–TrkB axis, and thus inhibits the formation of malignant synapses.

[0045] Figure 16A) Flowchart demonstrating the effect of HPAC-siNTRK2@Gel combined with chemoradiotherapy on inhibiting invasion and prolonging survival in a patient-derived GBM model; B) Flowchart of the experimental process for evaluating HPAC_siNTRK2@Gel and its combination with radiotherapy and temozolomide (TMZ) in an NSG mouse orthotopic transplantation model derived from orthotopic tumors formed from patient-derived GSCs labeled with luciferase and tdTomato; C) Representative bioluminescence imaging images of NSG mice carrying intracranial luciferase⁺ GBM tumors under different treatment conditions, n = 3 mice per group; D) Evaluation of tumor burden changes in the shown intracranial orthotopic xenograft models over a 60-day observation period using longitudinal bioluminescence imaging, n = 3 biological replicates per group; E) Kaplan-Meier survival curves of the shown intracranial orthotopic xenograft tumor models, with median survival indicated. Survival differences were assessed using the Mantel-Cox test, n = 7 biological replicates per group; G0 Representative images of invasive tumor cell clusters at the tumor-brain interface of GBM xenografts from 147 patients, with tdTomato⁺ tumor cells in red and nuclei counterstained with DAPI (blue); F, quantitative analysis of the number and total area (μm²) of detached tumor cell clusters in the tumor-brain interface region (n = 9), with each group of samples independently taken from 3 mice; G, representative images of Ki67 (green) and cleaved caspase-3 (red) immunofluorescence labeling at the tumor-brain interface; H, quantitative analysis of Ki67⁺ proliferating cells, cleaved caspase-3⁺ apoptotic cells, and the co-localized area (μm²) of Ki67 and cleaved caspase-3 in each group (n = 9). Nine slides were obtained from three biologically independent animals; I, representative images of Synapsin-1 (green) and PSD95 (magenta) expression at the tumor-brain interface; J, quantitative analysis of Synapsin-1 and PSD95 fluorescence intensity in the tumor region, normalized to the mean intensity of the radiotherapy group, with n = 9 slides per group from three biologically independent animals; data in C, F, H, and J are expressed as mean ± standard deviation (sd); statistical analysis was performed using one-way ANOVA with Tukey's post-hoc test (F, H), or unpaired two-sided Student's t test (J) (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001);

[0046] Figure 17This study demonstrates the antitumor effect of HPAC-siNTRK2@Gel-enhanced radiotherapy on a patient-derived classical glioblastoma model. A) Representative bioluminescence images of NSG mice with patient-derived intracranial tumors (G0 927 mice) before treatment, after radiotherapy alone, and after 35 days of treatment with HPAC-siNTRK2@Gel combined with radiotherapy. B) Quantitative analysis of bioluminescence intensity at different time points in the radiotherapy group and the radiotherapy + HPAC-siNTRK2@Gel group. Data are expressed as mean ± standard deviation (n=3 mice per group). Statistical significance was assessed using two-way ANOVA and Tukey's post-hoc test. Detailed Implementation

[0047] To better illustrate the purpose, technical solution, and advantages of this invention, the invention will be further described below with reference to specific drawings and embodiments. Unless otherwise specified, the experimental methods used in the embodiments are conventional methods, and the materials and reagents used are commercially available unless otherwise specified.

[0048] Example 1: Formation of malignant synapses at the GBM tumor-brain interface: a potential target for local therapy

[0049] 1. Multiplex fluorescence immunohistochemical staining and synaptic quantitative analysis

[0050] All human glioma tissue samples used in this study were approved by the Ethics Committee of Sun Yat-sen University Cancer Center. All enrolled subjects were adult glioma patients undergoing neurosurgical resection, and all had signed written informed consent forms before surgery. Surgically resected specimens included the central tumor region (tumor core) and the peripheral tumor region located at the junction of the tumor and brain tissue, defined based on its anatomical relationship with adjacent non-tumor brain tissue. All samples were anonymized in subsequent experimental processing; basic clinical and pathological information of the patients is provided in Appendix Table S2. Resected tissue was fixed in 10% neutral formalin and embedded in paraffin to prepare serial sections for multiplex immunohistochemical staining. Multiplex fluorescent labeling was performed using the Opal™ 4-Color Manual IHC Kit (Akoya Biosciences) to achieve spatial co-localization analysis of tumor cell markers and synaptic-related markers. Sections were first dewaxed in xylene and then rehydrated stepwise with graded ethanol. Subsequently, heat-induced antigen retrieval was performed before each round of antibody incubation.

[0051] The primary antibodies, working concentrations, and corresponding Opal fluorescent dyes used in this study are as follows, and staining was performed in the following order: Nestin (MAB1259, 1:1000, R&D Systems) paired with Opal 620; NeuN (MAB377B, 1:150, MilliporeSigma) paired with Opal 520; PSD95 (#3409, 1:400, Cell Signaling Technology) paired with Opal 570; Synapsin-1 (AB1543P, 1:1000, MilliporeSigma) paired with Opal 650. Signal amplification and antibody stripping were strictly performed according to the kit instructions between each round of staining to ensure the specificity and stability of the multiplex fluorescent labeling. After staining, the tissue sections were scanned and subjected to preliminary signal quantification analysis using the Vectra Polaris™ fully automated quantitative pathological imaging system (AkoyaBiosciences). To conduct regional analysis and synaptic quantification, serial hematoxylin and eosin (H&E) stained sections were independently interpreted by qualified neuropathologists without knowledge of the experimental grouping information to clarify anatomical regions, including the tumor core, tumor-brain interface, distal white matter, and distal cortex. Based on this, multiplex fluorescence images of corresponding anatomical regions were selected and high-resolution imaging was performed using a Nikon AXR confocal microscope. Quantitative analysis of synaptic structures was performed using the Synapse Counter plugin in ImageJ software. Statistical analysis of the colocalization signals of the presynaptic marker Synapsin-1 and the postsynaptic marker PSD95 was conducted to assess the spatial distribution characteristics of tumor-neuron synaptic structures in different anatomical regions.

[0052] 2. Gene set construction and single-sample gene set enrichment analysis (ssGSEA)

[0053] The transcriptomic data used in this study were obtained from the Chinese Glioma Genome Atlas (CGGA). After processing and preprocessing, the obtained transcriptomic data were used for subsequent gene set enrichment analysis to assess the transcriptional characteristics of signaling pathways related to neuronal function in gliomas.

[0054] Based on previous literature reports and the molecular functions related to neural synapses, this study constructed three gene sets closely related to neuronal signal transduction: (1) a gene set related to synaptic structure, including HOMER1, HOMER2, SHANK3, TTY12, CNTN1, CNTN2, NTRK2, NTRK3, and GJA1; (2) a gene set related to calcium channels, including CACNA1A, CACNA1C, CACNA1D, CACNA1E, CACNA1F, CACNA1G, CACNA2D1, CACNA2D2, CACNA2D3, CACNB1, CACNB2, CACNB3, CACNB4, and CACNG8; and (3) a gene set related to glutamate receptors, including GRIA1, GRIA2, GRIA3, GRIA4, GRIK1, GRIN2C, GRM2, and GRM7. Single-sample gene set enrichment analysis (ssGSEA) was performed using the GSVA software package in R. Enrichment scores were calculated for each sample to reflect the transcriptional activity levels of related signaling pathways at a single sample level. In subsequent grouping analysis, samples were divided into high-expression and low-expression groups based on the expression levels of TrkB (NTRK2), NLGN3, TSP1, CHRM3, and AMPAR-related genes, using the median of each gene in the entire sample as the boundary. Differences in gene set enrichment scores between different expression groups were statistically analyzed using the Wilcoxon rank-sum test; a p-value < 0.05 was considered statistically significant.

[0055] To investigate the formation of synaptic connections between glioblastoma (GBM) cells and neurons, we analyzed clinical specimens from patients who underwent en-bloc GBM surgery. Multiplex immunofluorescence staining was used to detect tumor stem markers (Nestin), neuronal cell body markers (NeuN), presynaptic markers (Synapsin-1), and postsynaptic markers (PSD95), and comparisons were made in different anatomically defined regions, including the necrotic core, tumor-brain interface, distal cortex, and distal white matter. Figure 1 A–1C). The results showed that Nestin was significantly enriched in tumor cells at the tumor-brain interface region, while Nestin signaling was extremely weak in cells within the necrotic core. Figure 1 A, 1C, 1D; Figure 2NeuN-positive neuronal cell bodies are abundant in the distal cortex, but sparse in the tumor-brain interface region, and are almost entirely absent in the distal white matter. Figure 1 A, 1C, 1D; Figure 2 Despite the low number of neuronal cell bodies at the tumor-brain interface, we still detected significant Synapsin-1 and PSD95 signals. Quantitative analysis showed that the fluorescence intensity of both Synapsin-1 and PSD95 was significantly increased in the tumor-brain interface region compared to the necrotic core. Figure 1 D).

[0056] Further colocalization analysis of Synapsin-1 and PSD95 puncta signals revealed that the number of inferred excitatory synaptic-like structures in the tumor-brain interface region was significantly higher than that in the necrotic core. Figure 1 E). Although the total number of colocalized punctate structures at the tumor-brain interface is lower than that in the distal cortex, their average punctate size is comparable to that in the distal cortex. Figure 1 F). The above quantitative characteristics were consistently validated in a sample of 6 independent GBM patients ( Figure 3 A-3D).

[0057] To further characterize the distribution of synaptic markers in the tumor periphery, we combined H&E staining with multiplex immunofluorescence staining analysis to additionally examine the anatomically defined tumor parenchyma, infiltrative margins, and distal cortex. Figure 3 E-3H). In these regions, the Synapsin-1 and PSD95 signals at the invasion margin were significantly lower than those in the tumor parenchyma and distal cortex. Figure 3 E-3H). In addition... Figure 3 I-3K presents a GBM specimen with tumor extending into the cortex, suggesting that physiological neurons—neuronal synapses—can coexist with tumor-associated synaptic-like punctate structures simultaneously within the cortical tissue.

[0058] To explore molecular signatures associated with tumor-neuron interactions, we further analyzed publicly available single-cell RNA sequencing (scRNA-seq) data, comparing the peritumoral brain (PTB) with the tumor core (TC) region. Figure 1G). Compared to TC tumor cells, GBM cells in the PTB region showed significantly upregulated expression of synapse-related genes, calcium channels, and glutamate receptors, supporting their enhanced synaptic connectivity with neurons. This result is consistent with previous preclinical models reporting synaptic interactions between GBM cells and neurons at the tumor-brain interface. Given that neuronal paracrine signaling activation via the BDNF-TrkB axis can significantly enhance malignant synapse formation between GBM cells and neurons, we further evaluated NTRK2 (TrkB) expression and found it to be significantly elevated in PTB tumor cells (G). Figure 1 G). Consistent with this, broader transcriptomic analyses also support the association between high TrkB expression and programmed enrichment of synapse-related genes. Figure 3 L-3Q).

[0059] Given that over 85% of GBM recurrences occur in the area adjacent to the surgical cavity, and that residual tumor cells in this area are often located at the tumor-brain interface, the above results suggest that the tumor-neuronal synapse may constitute a potential therapeutic target.

[0060] Example 2: HPAC nanocarrier targets GBM cells and achieves GSH-triggered siNTRK2 release.

[0061] 1. Synthesis and Characterization of HPAA-PEG-cRGD (HPAC)

[0062] The synthesis of HPAA-PEG-cRGD (HPAC) involves three steps: preparation of HPAA, construction of HPAA-PEG, and further modification of cRGD.

[0063] First, HPAA was synthesized via Michael addition polymerization. 0.64 g of N,N′-bis(acryloyl)cysteine ​​(CBA) was dissolved in 10 mL of a methanol / water mixture (3:1, v / v), and CaCl2 was added to bring the final concentration to 200 mM. The reaction system was placed in a 50 mL two-necked flask and magnetically stirred under nitrogen protection until completely dissolved. Then, 164 μL of N-(2-aminoethyl)piperazine was slowly added dropwise to the reaction system using a syringe, and the reaction temperature was maintained at 50 °C using a circulating water bath for 36 h. To block the terminal groups of the polymer, 300 μL of piperazine was added to the system, and the reaction was continued at 50 °C for another 12 h. After the reaction was complete, the pH of the system was adjusted to 3–4 using hydrochloric acid solution. The mixture was then dialyzed against deionized water for 2 days using a dialysis bag with a molecular weight cutoff of 10,000 to remove small molecule impurities. Finally, the mixture was freeze-dried to obtain a light yellow solid, HPAA.

[0064] HPAA-PEG was then synthesized. 200 mg of Mal-amido-PEG1-C2-NHS was dissolved in PBS buffer at pH 7.4 and mixed with HPAA at a molar ratio of PEG to HPAA of 1:10. The reaction was carried out under light-protected conditions at 25°C with stirring for 2 hours. After the reaction was complete, the mixture was purified by centrifugation at 4°C and 4000 rpm for 20 minutes using an ultrafiltration tube with a molecular weight cutoff of 10000. This process was repeated 1-2 times to remove unreacted substances. The purified HPAA-PEG was collected and freeze-dried.

[0065] Based on this, cRGD was modified onto HPAA-PEG via an addition reaction to obtain the final product HPAC. 50 mg of HPAA-PEG was dissolved in PBS buffer (pH 6.5–7.0), and 2 mg of HS-PEG-cRGD was also dissolved in PBS buffer. These solutions were slowly added dropwise to the HPAA-PEG solution over 30 min. The reaction system was stirred at room temperature for 48 h in the dark. After the reaction was complete, the product was dialyzed using a dialysis bag with a molecular weight cutoff of 10,000 to remove unreacted HS-PEG-cRGD and small molecule impurities. Finally, HPAA-PEG-cRGD (HPAC) was obtained by freeze-drying.

[0066] The chemical structures of the obtained HPAA, HPAA-PEG, and HPAC were characterized by ¹H nuclear magnetic resonance (¹H NMR) spectroscopy in D₂O solvent at a test temperature of 25 °C. The responsiveness of HPAA to reduced glutathione (GSH) was analyzed by gel permeation chromatography (GPC). The morphology of the polymers and their complexes was observed by transmission electron microscopy (TEM), and the particle size distribution and zeta potential of HPAC were determined using dynamic light scattering.

[0067] 2. Agarose gel electrophoresis analysis

[0068] To evaluate the binding affinity between HPAC and siRNA, HPAC / siRNA complexes were prepared at different mass ratios: 0.1:1, 1:1, 5:1, 10:1, 20:1, 40:1, 60:1, 80:1, and 120:1. After mixing HPAC and siRNA, the mixtures were incubated at 25°C for 30 min to ensure sufficient interaction. Following incubation, the resulting HPAC / siRNA complexes and free siRNA control samples were mixed with RNA loading buffer and then separated by electrophoresis on a 5% agarose gel in a DNase / RNase-free 1×TAE (tris-acetate-EDTA) buffer system at 150V for 30 min. After electrophoresis, the mixtures were analyzed using SYBR Green. TMThe gel was stained with Gold nucleic acid gel dye, and the gel was imaged and recorded using a gel imaging system.

[0069] 3. Preparation and characterization of SM-102-siRNA lipid nanoparticles (LNP)

[0070] SM-102-siRNA lipid nanoparticles (LNPs) were prepared using a microfluidic mixing method. Ionizable lipids SM-102, DSPC, cholesterol, and DMG-PEG2000 were dissolved in anhydrous ethanol at a molar ratio of 50:10:38.5:1.5 to prepare an organic phase solution with a total lipid concentration of 14.33 mM. The aqueous phase containing siRNA was prepared by dissolving oligonucleotides in 25 mM sodium acetate buffer (pH 4.0). Subsequently, the lipid organic phase and the siRNA aqueous phase were rapidly mixed in a microfluidic mixer at a total flow rate of 6 mL / min and an aqueous phase to organic phase volumetric flow rate ratio of 3:1 to form an SM-102-siRNA lipid nanoparticle suspension. The prepared LNP suspension was then subjected to buffer replacement via ultrafiltration to remove residual ethanol. The specific method involved using ultrafiltration centrifuge tubes with a molecular weight cutoff of 100 kDa, centrifuging at 3600 × g for 30 min at 4–6 °C, repeated three times, with the buffer solution replaced with PBS buffer at pH 7.4. The morphology of the obtained LNPs was observed using transmission electron microscopy, and their hydrated particle size distribution and Zeta potential were determined using dynamic light scattering.

[0071] 4. Glioblastoma cells derived from patients

[0072] This study used four patient-derived glioblastoma stem-like cells (GSCs) to evaluate the therapeutic efficacy and robustness of HPAC-siNTRK2@Gel under different molecular subtypes, including G0 147, G0 927, GSC1203, and GSC2. G0 147 is a mesenchymal (MES) GSC cell line, derived from Professor Zhang Nu's research group at the First Affiliated Hospital of Sun Yat-sen University and established in previous studies. G0 927, GSC1203, and GSC2 were all derived from fresh tumor specimens from glioblastoma patients who underwent surgical resection at Sun Yat-sen University Cancer Center and were established in this study.

[0073] Tumor tissue was processed immediately after surgical resection. Resected specimens were transported in sterile PBS containing amphotericin B and gentamicin, mechanically minced, and digested with 0.05% Trypsin-EDTA containing 0.2 mM EDTA at 37°C for 30–45 min. The digested cell suspension was filtered through a 70 μm nylon filter, centrifuged at 800 rpm for 5 min at 4°C, the supernatant was discarded, and the cells were resuspended in serum-free DMEM / F12 medium supplemented with epidermal growth factor (EGF, 20 ng / mL), basic fibroblast growth factor (bFGF, 10 ng / mL), and 2% B27 additive. Cells were cultured in suspension in low-attachment plates to form tumor spheroids and were periodically passaged into single-cell suspensions using StemPro Accutase. Cell viability was assessed by AO / PI staining, and the viability was consistently maintained above 80%. All cells were cultured at 37°C and 5% CO2 saturated humidity.

[0074] To clarify the molecular subtyping of each GSC cell line, whole-cell transcriptome sequencing (RNA-seq) analysis was performed on all GSC samples used in this study. Based on the characteristic gene set method, glioblastoma samples were classified into three molecular subtypes: Classical, Mesenchymal, and Proneural. RNA-seq data were quantified in FPKM form and log2 transformed. For each molecular subtype, characteristic genes detectable in the expression matrix were screened, and single-sample gene set enrichment analysis (ssGSEA) was used to calculate the subtype score. The obtained scores were Z-score normalized between samples. When the difference between a subtype score and the second-highest score was greater than 0.5, the sample was classified as that subtype; otherwise, it was defined as mixed. The subtype score results were visualized using a heatmap.

[0075] In in vivo xenograft experiments, GSC cells were transduced with lentiviruses to co-express firefly luciferase and tdTomato fluorescent protein, and a stable double-labeled cell line was established through puromycin selection (2 μg / mL, 5–7 days). The obtained cells were then used to construct an orthotopic patient-derived xenograft (PDX) mouse model for in vivo bioluminescence imaging and histological analysis. Representative GSC models of different molecular subtypes, including G0147 (mesenchymal), G0927 (classical), GSC1203 (mixed), and GSC2 (anterior neural), were ultimately selected to systematically evaluate the antitumor efficacy of HPAC-siNTRK2@Gel hydrogel therapy.

[0076] 5. Isolation and culture of primary mouse astrocytes

[0077] Primary astrocytes were obtained from the brain tissue of newborn C57 BL / 6 mice (0-2 days after birth). Immediately after collection, the brain tissue was dissected and digested with 0.25% trypsin at 37°C for 30 min, followed by mechanical pipetting to obtain a single-cell suspension. The cell suspension was filtered through a 70 μm nylon mesh and seeded into uncoated culture dishes in DMEM / F12 medium. After 10-14 days of culture, the mixed glial cells reached a confluence state, and then astrocytes were isolated using a mild trypsin digestion method. Specifically, a short-term digestion with 0.05% Trypsin-EDTA containing 0.2 mM EDTA was used to selectively isolate astrocytes. The isolated astrocytes were further expanded and cultured in DMEM / F12 medium supplemented with 10% fetal bovine serum and antibiotics.

[0078] GL261 Glioma Cell Line: This study used the mouse glioblastoma cell line GL261 and constructed and applied three cell forms according to different experimental needs: unlabeled GL261 cells, luciferase-labeled GL261 cells (GL261-luc), and green fluorescent protein-labeled GL261 cells (GL261-GFP). Unlabeled GL261 cells were cultured in DMEM medium containing 10% fetal bovine serum and 0.1 mg / mL Primocin at 37°C and 5% CO2, and mycoplasma contamination was detected regularly. GL261-luc cells stably expressed firefly luciferase in GL261 cells via lentiviral transduction for in vivo bioluminescence imaging experiments. After cell labeling, the luminescence signal was verified using an in vivo imaging system to confirm stable luciferase expression. GL261-GFP cells were subsequently used for patch-seq experiments, combining whole-cell patch-clamp electrophysiological recording with single-cell RNA sequencing analysis. This cell line was constructed by stably expressing the green fluorescent protein CopGFP in GL261 cells via lentiviral transduction. After lentiviral transduction, both GL261-luc and GL261-GFP cells were selected using puromycin at 2 μg / mL for 5–7 days to establish a stable cell population expressing the marker gene. All GL261 cell lines were maintained and expanded under the same culture conditions for subsequent in vivo and in vitro functional experiments.

[0079] 6. CCK8 drug cytotoxicity detection

[0080] The cytotoxicity of HPAC was evaluated using the CCK-8 assay, with HPAA, PEI-25K (23966, Polysciences, USA), and SM-102 LNP serving as control groups for comparison. GL261 cells were used as model cells, seeded in 96-well plates at 3000 cells per well, and cultured at 37°C and 5% CO2 for 24 h. After culture, the original medium was discarded and replaced with DMEM medium containing different concentrations of HPAC, HPAA, PEI-25K, or LNP, at a volume of 100 μL per well. Cells treated with only medium served as a blank control group. Cells were then co-cultured at 37°C for another 24 h. After treatment, cells in each well were gently washed once with PBS buffer, and then 100 μL of DMEM medium and 10 μL of CCK-8 reagent were added to each well, incubated at 37°C in the dark for 1 h. After incubation, the absorbance of each well was measured at 450 nm using a microplate reader.

[0081] Cell viability is calculated using the following formula:

[0082] Cell viability (%) = (A - B) / (C - B) × 100

[0083] Where A represents the absorbance value of the HPAC, HPAA, PEI-25K or LNP treatment group, B represents the absorbance value of the blank control group (containing only culture medium), and C represents the absorbance value of the untreated control group. All experiments were independently repeated 3 times.

[0084] 7. Cell uptake, transfection, and intracellular release analysis

[0085] Cellular uptake assay: To observe the intracellular uptake of HPAC, HPAC was fluorescently labeled. Specifically, HPAC and Cy5 were dissolved in PBS buffer and reacted with magnetic stirring at room temperature for 2 hours. Unbound Cy5 molecules were then removed by dialysis, and finally, Cy5-labeled HPAC was obtained by freeze-drying. GL261 cells were seeded in 24-well plates at a density of 5 × 10⁶ cells per well. 4 Cells were cultured at 37°C and 5% CO2 to approximately 90% confluence. Cy5-labeled HPAC was then added to each well to a final concentration of 1 mg / mL, and incubated for 0.5 h, 2 h, 4 h, and 8 h, respectively. After incubation, cells were gently washed three times with PBS buffer to remove untaken material, then gently digested and collected, and resuspended in PBS buffer. Intracellular fluorescence signals were detected by flow cytometry to quantitatively analyze HPAC uptake levels. Flow cytometry data were analyzed using FlowJo software.

[0086] siRNA transfection efficiency analysis: GL261 cells were seeded in confocal culture dishes, with 5 × 10⁶ cells per dish. 4 Cells were cultured at 37°C and 5% CO2 until adherence, then treated with serum-free Opti-MEM medium for 2 h. Subsequently, HPAA or HPAC was mixed with FAM-labeled siRNA at different mass ratios (20:1–140:1) in PBS buffer and incubated at room temperature for 30 min to form a complex. PEI-25K and SM-102 LNP were used as positive controls, and free siRNA as a negative control. The complex was added to cell culture dishes and incubated at 37°C for another 24 h. After treatment, cells were washed with PBS buffer, and the nuclei were stained with Hoechst 33342. After staining, confocal laser scanning microscopy (Nikon AXR) was used to image the cells for qualitative observation of siRNA uptake and transfection. After confocal imaging, cells were collected and resuspended in PBS buffer, and intracellular FAM fluorescence signals were detected by flow cytometry. Flow cytometry data were analyzed using FlowJo software, and the siRNA transfection efficiency was quantitatively assessed by statistically analyzing the proportion of FAM-positive cells.

[0087] GSH-responsive intracellular siRNA release analysis: To analyze the GSH-responsive release behavior of the HPAC / siRNA complex in cells, a complex of HPAC labeled with Cy5 and siRNA labeled with FAM was prepared using the same method as described in the previous experiments. The complex was then applied to primary mouse astrocytes, patient-derived primary glioblastoma cells, and GL261 cells, respectively. All cell types were seeded in confocal culture dishes at a density of 5 × 10⁶ cells per dish. 4 Cells were serum-free after adhesion and transfected according to a protocol optimized for GL261 cells. After incubation at 37°C for 24 h, cells were gently washed with PBS buffer and then stained with Hoechst 33342. Following staining, cells were imaged using a confocal laser scanning microscope, acquiring Hoechst, FAM, and Cy5 fluorescence signals at excitation wavelengths of 405 nm, 488 nm, and 555 nm. The acquired confocal images were analyzed using ImageJ software. The co-localization of FAM-siRNA and Cy5-HPAC signals was quantitatively assessed to reflect the release of siRNA from the GSH-enriched intracellular environment.

[0088] Analysis of the effect of siRNA loading on cRGD-mediated cellular uptake: To analyze the effect of siRNA loading on cRGD-mediated cellular uptake behavior, HPAA and HPAC were labeled with Cy5 fluorescence according to the aforementioned method. For the siRNA-loaded system, under the same conditions as the transfection experiment, Cy5-labeled HPAC was combined with siRNA to form a complex, thus preparing the HPAC-Cy5 / siRNA complex. GL261 glioma cells and primary astrocytes were seeded into 24-well plates, with 5 × 10⁶ cells per well. 4 Cells were cultured at 37°C and 5% CO2 to approximately 80%–90% confluence. Subsequently, HPAA-Cy5, HPAC-Cy5, or HPAC-Cy5 / siRNA were added to the cells, with consistent polymer concentrations across all treatment groups. Cells were incubated at 37°C for another 24 hours. After incubation, cells were washed three times with PBS buffer to remove unbound material, then mildly trypsinized and collected, resuspended in PBS buffer. Intracellular Cy5 fluorescence signal intensity was detected by flow cytometry, and the data were analyzed using FlowJo software. The average fluorescence intensity was used as an indicator to quantitatively compare the cellular uptake levels of different treatment systems in GL261 cells and primary astrocytes.

[0089] 8. HPAC-mediated in vitro NTRK2 gene silencing experiment

[0090] GL261 cells were seeded in 6-well plates at a density of 1 × 10⁶ cells per well. 5 Cells were cultured at 37°C and 5% CO2 for 24 h. Subsequently, cells were treated with free siRNA or siRNA delivered by different vectors, including HPAA, HPAC, SM-102 LNP, and PEI-25K. The amount of NTRK2 siRNA used in each treatment group was consistent, 2 μg per well, with the same polymer dosage used in the HPAA and HPAC delivery systems. Cells were cultured for another 48 h after treatment, followed by trypsin digestion and cell collection for protein extraction. The total protein concentration was determined using a BCA protein quantification kit (P0012, Beyotime, China). Equal amounts of protein samples were separated by SDS-PAGE electrophoresis and transferred to PVDF membranes for Western blot analysis according to standard procedures. The transferred PVDF membranes were incubated with the following primary antibodies: TrkB antibody (80E3, 1:1000, 4603, Cell Signaling Technology, USA) and phosphorylated TrkB antibody (pTyr). 5 ¹ 5The following antibodies were used: p44 / 42 MAPK (Erk1 / 2) antibody (Thr202 / Tyr204, 1:1000, 9102S, Cell Signaling Technology, USA), p44 / 42 MAPK (Erk1 / 2) antibody (Thr202 / Tyr204, 1:1000, 4370, Cell Signaling Technology, USA), and GAPDH antibody (1:10000, ab181602, Abcam, United Kingdom), with GAPDH serving as an internal control. Subsequently, HRP-labeled goat anti-rabbit secondary antibody (1:5000, ab6721, Abcam, United Kingdom) was added for incubation. Immunoblot signals were developed using a chemiluminescence detection kit (K1129, APExBIO, USA) and imaged and recorded using a ChemiDoc imaging system.

[0091] The NTRK2 small interfering RNAs (siRNAs) used in this study included three mouse-derived NTRK2 siRNAs (siNTRK2-1, siNTRK2-2, and siNTRK2-3), one human NTRK2 siRNA (hsiNTRK2), a negative control siRNA, and a fluorescently labeled FAM-siRNA, all synthesized by Qingke Biotechnology Co., Ltd. The FAM fluorescent group was linked to the oligonucleotide molecule via a phosphate ester bond during siRNA synthesis; this labeling was introduced during the synthesis stage using a phosphoramide chemical method. Detailed sequence information for all mouse and human NTRK2 siRNAs used in this study is shown in Table 1.

[0092] Table 1

[0093] Gene Sense strand Antisense strand murine siNTRK2-1 GCUCCUUAAGGAUAACGAA UUCGUUAUCCUUAAGGAGC murine siNTRK2-2 GGAGUCGACUACGAGACAA UUGUCUCGUAGUCGACUCC murine siNTRK2-3 GCGCUUCAGUGGUUCUACA UGUAGAACCACUGAAGCGC human siNTRK2 GGUUAGAAAUCAUCAACGA UCGUUGAUGAUUUCUAACC

[0094] We first synthesized the HPAA polymer backbone via a Michael addition reaction of N,N′-bis(acryloyl)cystamine and aminoethylpiperazine. Then, we used the abundant amino groups on HPAA to perform an amidation reaction with Mal-amido-PEG1-C2-NHS to obtain HPAA-PEG-MAL. Further, we performed a Michael addition reaction between the maleimide group of HPAA-PEG-C2-NHS and the thiol-containing HS-PEG-CRGD to obtain HPAA-PEG-CRGD (HPAC). The cRGD group can achieve tumor targeting by recognizing integrin αvβ3 on the surface of GBM cells. Finally, we combined siNTRK2 with HPAC to form the HPAC-siNTRK2 nanocomposite (…). Figure 4 A).

[0095] 1H nuclear magnetic resonance (1H NMR) confirmed the successful synthesis of HPAC. Figure 4 B). Dynamic light scattering (DLS) revealed that the hydrated particle size of HPAC nanoparticles under aqueous conditions was approximately 150 nm. Transmission electron microscopy (TEM) primarily reflected the core size in the dehydrated state (different from the hydrated state measured by DLS), revealing regularly shaped spherical nanostructures with relatively smaller apparent diameters. Figure 4 C, D). Zeta potential analysis showed that HPAC had a positive surface charge; its potential decreased after complexing with siRNA, and further incubation with 10 mM GSH resulted in a continued decrease in potential, suggesting charge neutralization and nanoparticle disassembly under reducing conditions. Figure 4 E). Agarose gel electrophoresis showed that when the mass ratio reached 20:1 (HPAC:siRNA) or higher, most siRNAs could be effectively adsorbed / bound by the HPAC vector. Figure 4 F). Furthermore, gel permeation chromatography (GPC) showed a clear elution peak shift after GSH treatment, indicating disulfide bond cleavage and polymer degradation. Figure 4 G).

[0096] Cytotoxicity assays in GL261 glioma cells showed that HPAC and its HPAA backbone exhibited low toxicity over a wide concentration range; in contrast, the commercially available cationic transfection reagent PEI-25k showed significant cytotoxicity at higher doses. Figure 4H). Flow cytometry analysis showed that Cy5-labeled HPAC nanocarriers exhibited time-dependent cellular uptake, with endocytosis reaching its peak after 4 h of incubation. Figure 5 A, 5B).

[0097] To evaluate intracellular siRNA delivery efficiency, we conjugated HPAC with FAM-labeled siRNA and compared it with non-targeted HPAA and PEI-25k under serum-containing conditions. Figure 4 Flow cytometry quantification results showed that, under various vector / siRNA quality ratios, HPAC had significantly higher transfection efficiency than HPAA. Figure 4 I). When the vector to siRNA mass ratio exceeds 80:1, the transfection efficiency no longer increases further. Figure 4 I). Representative flow cytometry histograms and their quantitative analysis both confirmed that HPAC has superior siRNA delivery capability ( Figure 4 J,K), and confocal laser scanning microscopy (CLSM) imaging results also provide support ( Figure 4 L). Consistent with its redox response design, pre-incubation of the HPAC / siRNA complex with GSH significantly reduced intracellular siRNA delivery efficiency, suggesting that siRNA was prematurely released before entering the cell. Figure 4 JL). Western blot further showed that HPAC-mediated siNTRK2 delivery effectively inhibited TrkB expression and its downstream pTrkB and pERK signaling (JL). Figure 4 M,N).

[0098] To compare the performance of HPAC with another commonly used nucleic acid delivery platform, we introduced SM-102 lipid nanoparticles (LNPs) in parallel for comparison. Figure 5 C-5K). LNPs also exhibited nanoscale characteristics under TEM / DLS and displayed different surface charge features in PBS (C-5K). Figure 5 C-5E, its cytotoxicity in GL261 cells is similar to that of HPAC ( Figure 5 F). However, under serum-containing conditions, HPAC achieved significantly higher siRNA transfection efficiency than LNPs, and TrkB knockdown was more pronounced (F). Figure 5 G-5K).

[0099] To determine whether siRNA surface loading affects cRGD-mediated tumor targeting, we compared cellular uptake of Cy5-labeled HPAC with and without siRNA. Flow cytometry showed no significant difference in Cy5 fluorescence intensity between HPAC_Cy5 and HPAC_Cy5 / siRNA in GL261 cells, while uptake of non-targeted HPAA_Cy5 was significantly reduced. Figure 5 L, 5N). In contrast, no enhanced uptake was observed among the formulations in primary astrocytes of newborn mice. Figure 5 M, 5N), indicating that the system has tumor-selective targeting characteristics.

[0100] CLSM imaging further confirmed that HPAC and siRNA can be effectively dissociated within GBM cells. Figure 4 O, P; Figure 5 O). Intracellular GSH levels in tumor cells (approximately 2–10 mM) are about three orders of magnitude higher than extracellular levels (approximately 2–10 μM), and can be several times higher than in normal cells. High-magnification imaging and line scanning analysis of GL261 cells, patient-derived primary GBM cells, and primary astrocytes showed that siRNA was selectively released in tumor cells, while its release was not significant in normal astrocytes. Figure 5 P).

[0101] In summary, these results demonstrate that the HPAC nanocarrier possesses efficient siRNA loading, integrin-mediated tumor targeting, and GSH-responsive intracellular release capabilities, enabling selective silencing of TrkB in GBM cells while reducing off-target delivery to normal glial cells.

[0102] Example 3: pH-responsive hydrogel for postoperative local treatment of GBM

[0103] 1. Preparation of Oxidized Hyaluronic Acid (OHA)

[0104] 5g of hyaluronic acid (HA) was weighed and dissolved in 200mL of deionized water. The solution was slowly stirred magnetically at room temperature until completely dissolved. Then, a 0.5M sodium periodate (NaIO4) solution (dissolved in 25mL of deionized water) was slowly added dropwise to the HA solution at 25°C in the dark to prevent photoinduced degradation of the oxidant. The reaction system was stirred continuously for 2 hours under the above conditions. After the reaction was completed, 1mL of glycerol was added to the system to terminate the oxidation reaction, and stirring was continued for 1 hour. The reaction solution was then placed in a dialysis bag with a molecular weight cutoff of 10,000 and dialyzed against deionized water for 3 days to remove unreacted reagents and small molecule byproducts. After dialysis, the resulting solution was freeze-dried to obtain oxidized hyaluronic acid (OHA) and stored at 4°C for later use.

[0105] 2. Macroscopic property characterization of hydrogels

[0106] The self-healing properties of the hydrogel were evaluated through macroscopic observation. Two circular hydrogels, approximately 2.2 cm in diameter, were prepared; one was stained with red dye, while the other remained colorless. Each hydrogel was cut in half, and one half (red) was bonded to the other (colorless) at 25°C. After standing for 30 minutes, the diffusion of dye towards the colorless hydrogel was observed, and the reconstituted hydrogel was gently stretched to assess the gel network reconstruction capability. The shape adaptability of the hydrogel was evaluated through a mold-forming experiment. Freshly prepared hydrogels were injected into silicone molds of different shapes, including circular, heart-shaped, triangular, and pentagonal. After a short standing period, the hydrogel's adaptation to the mold shape was observed. The injectability of the hydrogel was tested through an injection molding experiment. The hydrogel precursor solution was loaded into a syringe with an inner diameter of 0.45 mm, and the word "SYSUCC" was written on a flat surface by extrusion. The morphology of the formed hydrogel was recorded.

[0107] 3. Degradation and microstructure characterization of hydrogels

[0108] The final hydrogel formulation consisted of 3% OHA, 5% carboxymethyl chitosan (CM-chitosan), and 3% HPAC. The components were mixed sequentially in the above proportions and vortexed at 25°C for 10 seconds to induce rapid gelation. This formulation was used for subsequent pH-responsive degradation experiments and hydrogel microstructure analysis. To assess the Schiff base-mediated pH-dependent degradation behavior, approximately 1 mL of hydrogel was placed in a pre-weighed container, and the container mass was recorded as Wg. Then, 1.5 mL of PBS buffer at different pH values ​​(pH 7.4, 6.8, 6.5, or 6.0) was added to the container, and it was incubated in a rotary shaker at 37°C. At a predetermined time point, the supernatant PBS solution was removed, and the total mass of the container and remaining hydrogel was weighed and recorded as Wt. Fresh PBS of the same pH value was added, and incubation continued. The remaining mass percentage of the hydrogel was calculated using the following formula:

[0109] Remaining mass (%) = [(Wt - Wg) / (W0 - Wg)] × 100%;

[0110] Where W0 represents the total mass of the container and hydrogel in the initial gelation state. The microstructure of the hydrogel was observed using environmental scanning electron microscopy. After freeze-drying, the hydrogel sample underwent brittle fracture treatment, followed by sputtering a thin layer of gold-palladium alloy onto its surface. The internal morphology and pore structure were then imaged and analyzed using environmental scanning electron microscopy (ESEM, FEI Quanta FEG 450).

[0111] 4. Rheological property testing of hydrogels

[0112] To characterize the mechanical properties of different hydrogel formulations, five hydrogel samples with different compositions were prepared by adjusting the ratios of oxidized hyaluronic acid (OHA), carboxymethyl chitosan (CM-chitosan), and HPAC. Specific formulation ratios are described in the main text. Rheological testing of the hydrogels was performed using a KINEXOS Pro rheometer equipped with a 20 mm diameter parallel plate clamp. Approximately 0.5–1 mL of freshly mixed hydrogel sample was placed in the center of the lower plate of the rheometer, and the upper plate was slowly lowered to a gap height of 500 μm at 37°C. The gelation kinetics of the hydrogels were tested using time-scan mode. Under constant frequency of 10 Hz and strain of 1%, the changes in storage modulus (G') and loss modulus (G'') over time were continuously recorded. Shear thinning behavior was evaluated using frequency-scan mode. Under constant strain, the frequency was gradually increased from 0.1 Hz to 10 Hz, and the corresponding changes in rheological parameters were recorded. The mechanical stability of the hydrogel under different deformation conditions was tested using a strain scanning mode with a strain range of 0.1%–1000% to evaluate the structural stability of the hydrogel under progressively increasing deformation. The self-healing properties of the hydrogel were tested by alternating cycles of low strain (1%) and high strain (500%), recording the changes and recovery of G' and G'' at different strain stages. All rheological tests were independently repeated three times, and the results were statistically analyzed as averages.

[0113] 5. Evaluation of the in vitro proliferation and invasion capacity of drug-free hydrogel matrix glioma cells

[0114] 5.1 Transwell Transfer Experiment

[0115] GL261 cells were pre-cultured in control medium or OHA / CMC hydrogel extract, then collected and resuspended in serum-free medium. Cells were seeded in the upper chamber of a Transwell chamber, while the lower chamber contained complete medium with 10% fetal bovine serum as a chemokine source. After incubation at 37°C for 12 or 24 hours, unmigrated cells in the upper chamber were gently removed with a cotton swab. Cells that migrated to the membrane surface of the lower chamber were then fixed with 4% paraformaldehyde and stained with 0.5% crystal violet. After staining, the migrating cells were imaged under an inverted microscope. The number of migrating cells was quantitatively analyzed using ImageJ software.

[0116] 5.2 Scratch Healing Test

[0117] GL261 cells were seeded in culture plates and cultured at 37°C and 5% CO2 until a dense monolayer formed. Subsequently, uniform-width scratches were created on the surface of the cell monolayer using a sterile pipette tip, and the cells were washed with PBS buffer to remove detached cells. The treated cells were then added to either control medium or OHA / CMC hydrogel extraction medium for further culture. The scratched areas were imaged and recorded using an inverted microscope at 0h, 12h, and 24h after scratch formation. The scratch closure area was measured using ImageJ software, and cell migration ability was calculated as the reduction in scratch area relative to the 0h scratch area.

[0118] 5.3 Cell viability detection

[0119] To evaluate the effects of hydrogel extract on glioma cell proliferation and viability, GL261 cells were cultured in control medium or OHA / CMC hydrogel extract for 4 h, 12 h, 24 h, 36 h, and 48 h, respectively. Cell viability was assessed at each time point using the CCK-8 assay, following the kit instructions described above. After the reaction, absorbance values ​​of each well were measured using a microplate reader, and the data were normalized to a control group to compare changes in cell viability under different treatment conditions.

[0120] 6. Laboratory animals

[0121] All animal experiments in this study adhered to relevant animal ethics guidelines and were approved by the Animal Ethics Committee of Sun Yat-sen University Cancer Center (Approval No.: L102042024040B). The animals used included 6-8 week old male C57BL / 6J mice and 6-8 week old male NOD / ShiLtJGpt-Prkdc^em26Cd52 / Gpt mice, both purchased from Guangdong Jicui Pharmaceutical Biotechnology Co., Ltd. The animals were housed in an SPF-grade animal facility at Sun Yat-sen University Cancer Center, using a 12-hour light-dark cycle, and had free access to standard feed and water. The ambient temperature in the animal facility was maintained at 23℃, and the relative humidity was maintained at approximately 50%. After the follow-up period, the mice were euthanized by isoflurane overdose, supplemented by cervical dislocation to ensure complete death.

[0122] 6.1 Establishment of an intracranial GL261 glioma mouse model and treatment regimen

[0123] A 6-8 week old male C57BL / 6J mouse model of intracranial GL261 glioma was established. GL261 tumor cells expressing luciferase were prepared into a cell suspension, and each mouse was injected with 1×10⁻⁶ cells. 5Cells were suspended in 5 μL of PBS. Under 3% isoflurane inhalation anesthesia, the cells were injected into the right frontal lobe of mice using a stereotaxic instrument at a depth of 1.4 mm, 1 mm anterior to the anterior fontanelle and 0.5 mm lateral to the midline. Mice used for patch-seq experiments were injected with GFP-labeled GL261 cells under the same conditions to facilitate precise localization and identification of tumor cells. On day 8 post-inoculation, mice under isoflurane anesthesia underwent magnetic resonance imaging (Premier 3.0T, GE) to confirm the location and volume of the tumor. Based on the MRI localization results, craniotomy was performed under 20x stereomicroscopic magnification, and the tumor tissue was removed using fine forceps and a negative pressure suction device.

[0124] Immediately after tumor resection, 20 μL of the appropriate treatment agent was injected into the tumor cavity. Mice were randomly assigned to the following six treatment groups: PBS group, HPAC-siNTRK2 group, Gel group, HPAC-Negsi@Gel group, HP-siNTRK2@Gel group, and HPAC-siNTRK2@Gel group. Mice were monitored daily postoperatively, and were immediately euthanized when they showed a weight loss of more than 20% or obvious clinical abnormalities (including obvious kyphosis, lethargy, or significant decrease in activity).

[0125] For longitudinal follow-up evaluation, three mice from each group were randomly selected to undergo bioluminescence imaging and MRI at predetermined time points. Bioluminescence imaging was performed using the IVIS Spectrum imaging system on the day of treatment with the hydrogel or nanodelivery system, and on days 5, 12, 22, 30, and 45 post-surgery. Survival analysis included ten mice from each group, and weight change monitoring included six mice from each group.

[0126] On postoperative day 15, three mice from each group were randomly sacrificed, and brain tissue was collected for HE staining, Ki67 immunohistochemical staining, and TUNEL assay. Starting from postoperative day 15, behavioral assessments were performed on no fewer than 15 mice from each group to analyze anxiety-like behaviors (open field test, elevated cruciate maze), depression-like behaviors (tail suspension test, forced swimming test), and spatial and working memory functions (Y-maze test and water maze test).

[0127] 6.2 Establishment and Treatment Protocol of Patient-Derived GSC Xenograft Animal Model

[0128] A patient-derived intracranial xenograft model of glioblastoma was established using 6-8 week old male NOD / ShiLtJGpt-Prkdc^em26Cd52 / Gpt mice. The tumor cells used in the experiment were patient-derived glioma stem cells (GSCs) simultaneously expressing luciferase and tdTomato. GSCs were prepared into a cell suspension, and each mouse was injected with 1 × 10⁻⁶ cells.5 Cells were suspended in 5 μL PBS. Under isoflurane inhalation anesthesia, the cells were injected into the right frontal cortex of mice using a stereotaxic instrument at a depth of 1.4 mm, 1 mm anterior to the anterior fontanelle and 0.5 mm lateral to the midline. On day 8 post-tumor inoculation, mice under isoflurane anesthesia underwent magnetic resonance imaging to confirm successful tumor establishment and spatial location. Subsequently, microsurgery was performed on the mice under stereomicroscopy guidance to remove the tumor tissue. Immediately after tumor resection, 20 μL of HPAC-siNTRK2@Gel was injected into the tumor cavity. Tumor progression after treatment was monitored longitudinally using bioluminescence imaging; specific imaging time points are shown in the corresponding figure captions. In the survival analysis experiment, 7–12 mice were included in each group; the specific sample size was indicated in the corresponding figure captions for different experimental batches. To conduct mechanism-related analyses, some mice were sacrificed on day 12 post-treatment, and brain tissue was collected for patch-clamp electrophysiological recording and multiplex immunofluorescence staining to assess the formation of tumor-neuronal malignant synaptic structures.

[0129] 6.3 In vivo magnetic resonance imaging (MRI) detection in a mouse glioma model

[0130] Magnetic resonance imaging (MRI) was used to image a mouse model of intracranial glioma for tumor localization, preoperative and postoperative follow-up, and assessment of postoperative brain tissue structural changes. MRI scans were performed using a 3.0T MRI system (SignaPremier, GE Healthcare) equipped with a 4-channel, 5cm diameter small animal coil (Medcoil Healthcare). During imaging, mice were anesthetized with isoflurane under oxygen conditions, with each imaging session lasting approximately 15 minutes. The imaging sequence used was a T1-weighted FLAIR sequence with the following parameters: repetition time / echo time (TR / TE) of 1750 / 24 ms, planar resolution of 0.3 × 0.3 mm², matrix size of 192 × 192, and slice thickness of 0.5 mm. After identifying the tumor region using the localization sequence, 16 consecutive slices covering the tumor area were acquired. The acquisition time for a single T1-weighted sequence was 3 minutes and 55 seconds. Quantitative analysis of tumor and lesion volume was performed by manually delineating regions of interest (ROIs). ROIs were delineated layer by layer, and volume calculations were performed using ITK-SNAP software (version 4.0.1). All ROI delineation was performed independently by researchers unaware of the grouping information, and the volume measurement error between different operators was controlled within 5%. Based on continuous slice data, three-dimensional reconstructions were performed on postoperative tumor cavities and recurrent tumor areas to demonstrate their spatial distribution characteristics.

[0131] In the behavioral correlation analysis, the functional brain injury volume was defined as the sum of the postoperative tumor cavity volume and the recurrent tumor volume. This parameter was quantitatively calculated on day 15 postoperatively to characterize postoperative changes in brain tissue structure. The quantitative results were used for data interpretation in subsequent behavioral analyses.

[0132] 6.4 In vivo biosafety assessment

[0133] To evaluate the in vivo biosafety of different treatment systems under tumor-free conditions, healthy male C57BL / 6J mice were used for sham surgery experiments. Three mice were included in each group. During the experiment, mice underwent only a sham surgery procedure equivalent to craniotomy, without tumor cell implantation. Post-surgery, mice were randomly assigned to one of the following six treatment groups: PBS group, HPAC-siNTRK2 group, Gel group, HPAC-NegsiRNA@Gel group, HPAA-siNTRK2@Gel group, and HPAC-siNTRK2@Gel group. The treatment methods and administration procedures were consistent with those in the tumor model experiment. Mice were sacrificed on day 15 post-surgery, and peripheral blood samples and tissues from major organs, including the heart, liver, spleen, lungs, and kidneys, were collected. Serum samples were used for biochemical assays, including liver function indicators such as alanine aminotransferase, aspartate aminotransferase, albumin, and total bilirubin, and kidney function indicators such as blood urea nitrogen and creatinine. After the collected major organ tissues were fixed, embedded in paraffin, and sectioned, they were stained with hematoxylin and eosin (H&E) for histological observation to assess the histological changes of major organs under various treatment conditions.

[0134] 6.5 Postoperative assessment of neuroinflammatory response

[0135] To assess the neuroinflammatory response in the brain tissue surrounding the tumor cavity postoperatively, C57BL / 6J mice carrying GL261 orthotopic gliomas were used in the experiment. Mice were randomly divided into three groups: a control group (no treatment), a surgical group (PBS injected into the tumor cavity after tumor resection), and a hydrogel group (OHA+CMC+HPAC hydrogel injected into the tumor cavity without therapeutic siRNA after tumor resection). Mice were sacrificed on postoperative days 3, 7, and 28. Brain tissue was rapidly extracted after sacrifice at the designated time points, fixed in 4% paraformaldehyde at 4°C for 24–48 h, then routinely embedded in paraffin, and cut into serial sections approximately 5 μm thick. All subsequent analyses focused on the brain tissue region adjacent to the tumor cavity.

[0136] 6.6 GFAP / IBA1 dual immunofluorescence staining

[0137] Paraffin sections were dewaxed and hydrated, then subjected to heat-induced antigen retrieval, followed by blocking with a BSA-based blocking solution. The sections were incubated overnight at 4°C with a primary antibody mixture containing anti-IBA1 antibody (rabbit, GB113502, Servicebio, 1:300) and anti-GFAP antibody (mouse, GB12096, Servicebio, 1:500). After incubation, the sections were washed and incubated with CY3-labeled goat anti-rabbit IgG secondary antibody (GB21303, Servicebio, 1:300) and Alexa Fluor 488-labeled goat anti-mouse IgG secondary antibody (GB25301, Servicebio, 1:400), respectively. Subsequently, cell nuclei were counterstained with DAPI, and the sections were mounted with anti-fluorescence quenching mounting medium. All immunofluorescence images were acquired using a fluorescence microscope under the same imaging parameters.

[0138] 6.7 Immunohistochemical staining of inflammatory factors

[0139] To further assess the expression of inflammatory factors, immunohistochemical staining for IL-1β and TNF-α was performed on adjacent sections. After dewaxing, hydration, and antigen retrieval with citrate buffer, endogenous peroxidase activity was blocked with 3% hydrogen peroxide, followed by nonspecific blocking with 3% BSA. Subsequently, sections were incubated overnight at 4°C with either anti-IL-1β antibody (rabbit-derived, GB11113, Servicebio, 1:1000) or anti-TNF-α antibody (rabbit-derived, GB115702, Servicebio, 1:600). After washing, sections were incubated with HRP-labeled goat anti-rabbit IgG secondary antibody (Servicebio) and developed using the DAB staining system, followed by hematoxylin counterstaining of cell nuclei. After dehydration, clearing, and mounting, sections were used for bright-field microscopy.

[0140] 6.8 Image Acquisition and Quantitative Analysis

[0141] All immunofluorescence and immunohistochemical images were acquired under uniform acquisition parameters. ImageJ software was used to quantitatively analyze GFAP, IBA1, IL-1β, and TNF-α signals in the peritumoral region to compare changes in neuroinflammatory-related indicators under different treatment conditions.

[0142] 6.9 Postoperative neuronal toxicity assessment

[0143] To assess the neuronal toxicity of the brain tissue surrounding the tumor cavity after surgery, C57BL / 6J mice carrying GL261 orthotopic gliomas were used in the experiment. Mice were randomly divided into three groups: a control group (no treatment), a surgical group (PBS injected into the tumor cavity after tumor resection), and a hydrogel group (OHA+CMC+HPAC hydrogel injected into the tumor cavity without therapeutic siRNA after tumor resection). Brain tissue was collected on postoperative days 3, 7, and 28 for NeuN immunofluorescence staining analysis; Nissl staining was used to assess neuronal morphology and structural integrity. The analysis time point for the cortical region surrounding the tumor cavity was on postoperative days 3 and 28, and the analysis time point for the hippocampus was on postoperative day 14. After sacrifice, the brain tissue was rapidly removed, fixed with 4% paraformaldehyde, and then paraffin-embedded or frozen sectioned according to experimental requirements, with a section thickness of approximately 5 μm.

[0144] 6.10 Nissl staining

[0145] Nissl staining was used to observe the morphology and structural integrity of neurons. Paraffin sections were dewaxed with xylene and rehydrated to water using a gradient of ethanol; frozen sections were thawed and fixed. Sections were stained in Nissl staining solution for 2–5 min, followed by transient differentiation with 0.1% acetic acid, and thoroughly washed to terminate the differentiation process. After staining, sections were air-dried, cleared with xylene, and mounted with neutral resin. The stained sections were observed under a bright-field microscope, and representative images were acquired on postoperative days 3 and 28 in the peritumoral cortex, and on postoperative day 14 in the hippocampus for neuronal morphological analysis.

[0146] 6.11 NeuN Immunofluorescence Staining

[0147] To quantitatively assess neuronal viability, NeuN immunofluorescence staining was performed on paraffin sections. After dewaxing, hydration, and heat-induced antigen retrieval, the sections were blocked with a BSA-based blocking solution and then incubated overnight at 4°C with NeuN primary antibody (mouse-derived, GB120017, Servicebio, 1:100). After incubation, the sections were washed and incubated with AlexaFluor 488-labeled goat anti-mouse IgG secondary antibody (Servicebio) at room temperature in the dark. The nuclei were then counterstained with DAPI. The sections were mounted with antifluorescence-quenching mounting medium for later use.

[0148] 6.12 Image Acquisition and Quantitative Analysis

[0149] All immunofluorescence images were acquired using fluorescence microscopy under the same imaging parameters. ImageJ software was used to quantitatively analyze NeuN-positive neurons in the peritumoral cortex on postoperative days 3, 7, and 28 to compare neuronal survival under different treatment conditions.

[0150] 6.13 Assessment of in vivo hydrogel retention and degradation behavior

[0151] To evaluate the in vivo retention time, degradation kinetics, and responsiveness of hydrogels to the tumor recurrence microenvironment after a single intracavitary administration, FAM-labeled siRNA was used as a fluorescent tracer molecule for long-term in vivo tracking experiments. First, a GL261 orthotopic glioma model was established in C57BL / 6J mice. After tumor resection surgery, a hydrogel loaded with FAM-siRNA was injected into the tumor cavity in a single dose. By adjusting the concentrations of oxidized hyaluronic acid (OHA) and carboxymethyl chitosan (CMC) while keeping the HPAC content constant, three different hydrogel formulations were prepared: 5% OHA + 7% CMC + 3% HPAC, 3% OHA + 5% CMC + 3% HPAC, and 1% OHA + 3% CMC + 3% HPAC. All formulations were loaded with an equal amount of FAM-siRNA. These hydrogels were injected into the tumor cavity immediately after tumor resection. In addition, two tumor-free control groups were set up for comparative analysis: one group consisted of mice that underwent direct intracavitary hydrogel implantation without tumor cell inoculation; the other group underwent intracavitary hydrogel implantation first, followed by stereotactic injection of GL261 cells into the existing hydrogel on postoperative day 10 to simulate postoperative tumor recurrence. Mice were sacrificed on postoperative days 3, 10, 20, and 30, and whole brain tissue was collected for in vitro fluorescence imaging. The fluorescence signal of FAM-siRNA in the tumor cavity region was acquired using an IVIS imaging system under uniform parameter conditions. Quantitative analysis of the fluorescence intensity in the tumor cavity region was used to evaluate the retention time and degradation behavior of different hydrogel formulations in vivo.

[0152] To further verify the local retention of the hydrogel at the tissue level, brain tissue was harvested on postoperative day 30. After fixation with 4% paraformaldehyde, the tissue was frozen or paraffin-embedded as needed for the experiment. The sections were observed using a confocal laser scanning microscope to detect residual FAM-siRNA fluorescence signals within the tumor cavity, and the cell nuclei were counterstained using DAPI.

[0153] We aim to deliver siRNA to residual tumor cells around the surgical cavity using a synthesized HPAC nanocarrier. One of the main challenges of intraoperative cavity therapy is that locally administered formulations can be rapidly cleared by cerebrospinal fluid (CSF). Hydrogels are promising platforms for local drug delivery due to their good biocompatibility, injectability, adaptability to irregular cavities, and ability to achieve sustained and stimulus-responsive release. Considering the acidic extracellular microenvironment (pH approximately 6.2–6.9) created by GBM metabolic reprogramming, we designed a pH-responsive hydrogel that remains stable under physiological conditions but undergoes controlled disassembly in an acidic environment, thereby releasing the encapsulated HPAC / siRNA nanocomplex on demand. Figure 6 A).

[0154] This hydrogel is assembled from oxidized hyaluronic acid (OHA) and carboxymethyl chitosan (CMC), and forms a network structure with HPAC-siNTRK2. Its cross-linking mechanism mainly relies on dynamic imine bonds: the aldehyde groups on OHA react reversibly with the amino groups on CMC and HPAC. The individual solutions of OHA, CMC, and HPAC have low viscosity; however, rapid mixing of equal volumes can form a homogeneous gel within 10 seconds. Figure 6 B). Rheological tests confirmed successful gelation, as evidenced by the storage modulus (G′) consistently exceeding the loss modulus (G″) throughout the testing process. Figure 6 E). The resulting HPAC-siNTRK2@Gel can be smoothly injected through a 0.45 mm needle and maintains the shaped structure, thus adapting to complex postoperative cavity geometry. Figure 6 C, D). Oscillation amplitude scans show a clear yield point at high strain (approximately 110%), where G′ and G″ tend to coincide, indicating reversible failure of the network structure. Figure 6 F); shear rate-dependent viscosity tests further demonstrate that the gel exhibits significant shear-thinning properties. Figure 6 G). Furthermore, after HPAC was incorporated into the OHA / CMC matrix, both G′ and viscosity in the linear viscoelastic region increased, suggesting that the additional amine groups provided by HPAC increased the crosslinking density. Figure 6 E–G).

[0155] The dynamic imine network endows this hydrogel with excellent self-healing capabilities. At the macroscopic level, two severed pieces of gel can rapidly re-bond: dye diffusion across the interface can be observed within 30 minutes, and mechanical integrity is restored during tensile testing. Figure 6H). At the microscopic level, cyclic strain testing showed that G′ and G″ could recover rapidly and repeatedly after alternating loading of high and low strain, further confirming that the gel network has a highly efficient self-healing ability. Figure 6 I). These properties help ensure that the gel reliably fills the surgical cavity and maintains structural stability under postoperative mechanical disturbance.

[0156] To characterize the pH response, we monitored the swelling and degradation behavior of the hydrogel in PBS at different pH values. At pH 7.4, the gel underwent a brief initial swelling followed by a relatively stable mass; however, under acidic conditions, the gel mass gradually decreased in a pH-dependent manner. Figure 6 J). Scanning electron microscopy showed that the porous microstructure of the gel was significantly fragmented and collapsed in an acidic environment, while the network structure remained relatively intact at pH 7.4. Figure 6 (K). Notably, previous studies have suggested that lower extracellular pH in solid tumor cells is often associated with higher malignancy; consistent with this, our hydrogel degrades more rapidly at higher acidity levels, thus facilitating faster release of therapeutic siRNA. This pH-responsive behavior enables the hydrogel to degrade more quickly under tumor-associated acidic conditions.

[0157] This hydrogel also exhibits strong tissue adhesion, capable of supporting the weight of various isolated mouse organs (including brain, heart, liver, spleen, lungs, and kidneys). Figure 6 More importantly, we directly assessed its local biocompatibility in the brain. In the GL261 model, immunofluorescence analysis of the perioperative region showed that implantation of OHA / CMC+HPAC Gel did not induce sustained astrocyte or microglia activation. At 7 days post-operation, compared with the surgical control group alone, the gel-treated group had lower GFAP and IBA-1 positive signals, and these levels remained low at 28 days post-operation. Figure 6 The M–O result suggests that hydrogel implantation does not exacerbate postoperative neuroinflammatory response. Further analysis of neuroinflammatory markers also supports this conclusion. Figure 7 (A–D).

[0158] Nissl staining of the perioperative cortex showed a decrease in Nissl bodies on postoperative day 3, gradually recovering by day 28; and both the surgical group and the gel treatment group showed similar temporal trends. NeuN immunofluorescence results further indicated that the neuronal density in the gel treatment group did not show a sustained decrease compared to the surgical group at any time point; simultaneously, Nissl staining in the hippocampus on postoperative day 14 was comparable across all groups. Figure 7 In summary, intraoperative hydrogel implantation does not exacerbate surgery-related neuronal damage or cause additional neurotoxicity.

[0159] We further evaluated the systemic biocompatibility of the hydrogel formulation in GL261 orthotopic glioma-bearing mice. On day 15 after intraoperative administration, histological examinations were performed on major organs including the heart, liver, spleen, lungs, and kidneys. No significant tissue damage or pathological abnormalities were observed in any treatment group. Figure 8 A). Serum biochemical tests showed no significant differences in liver and kidney function-related indicators among the groups and were within the normal range. Figure 8 B). The above results suggest that intraoperative delivery of the hydrogel formulation did not cause detectable systemic toxicity.

[0160] After GL261 cells were exposed to drug-free OHA / CMC hydrogel extract, no changes in cell migration behavior were observed in Transwell and scratch healing assays at 12 h and 24 h. Figure 8 C–F); cell viability at each time point was also comparable to that of the control group. Figure 8 G). These results suggest that the OHA / CMC carrier matrix itself does not promote glioma cell migration or proliferation in vitro.

[0161] Finally, we used FAM-labeled siRNA to evaluate the retention and degradation behavior of the hydrogel in vivo after a single intraoperative dose. Figure 9 A). In practical applications, we need this gel to remain within the tumor cavity without degrading too quickly, and also to degrade and release the hyperbranched polymer HPAC carrying siNTRK2 in a timely manner during tumor recurrence. G4 and G5 simulate the tumor recurrence process, in which the gel is implanted in the brains of normal mice in advance; one group is not implanted (G4), and the other group is implanted on day 10 (G5). In vitro fluorescence imaging shows that hydrogel retention is closely related to the formulation: the optimized formulation (3% OHA + 5% CMC + 3% HPAC) shows a gradual decrease in fluorescence signal, but it can still be clearly detected 30 days after surgery. Figure 9 B, C). In contrast, the "harder" formulation (5% OHA + 7% CMC + 3% HPAC) showed minimal signal attenuation (G1), with almost no degradation after 30 days; while the "softer" formulation (1% OHA + 3% CMC + 3% HPAC) was almost completely degraded and invisible by around day 20, i.e., losing most of the fluorescence signal (G3). In the delayed tumor inoculation model, after GL261 cells were inoculated into the existing hydrogel, the fluorescence signal attenuated more rapidly (G5), indicating that the G5 gel rapidly responded to the acidic microenvironment caused by tumor proliferation and degraded; while the signal in the uninoculated control group remained generally stable (G4), suggesting that the hydrogel can respond to changes in the tumor-related microenvironment within a certain recurrence window. Figure 9B, C). Consistently, confocal imaging still showed residual FAM-siRNA signals in the resection cavity at 30 days post-surgery (B, C). Figure 9 D).

[0162] Therefore, a hydrogel containing 3% OHA, 5% CMC, and 3% HPAC was selected for subsequent animal experiments.

[0163] In summary, the HPAC-siNTRK2@Gel system combines injectability, shape adaptability, self-healing ability, pH-responsive disassembly, and strong tissue adhesion, and exhibits good neurobiocompatibility. The carrier matrix itself does not show a tumor-promoting effect. Furthermore, it has the characteristics of continuous retention and controllable degradation in vivo after a single dose, supporting its applicability as a drug delivery platform for postoperative GBM local treatment.

[0164] Example 4: Intraoperative HPAC-siNTRK2@Gel inhibits postoperative GBM recurrence and prolongs survival in various orthotopic glioma models.

[0165] 1. Tissue processing and analysis after HE, TUNEL and Ki67 staining

[0166] On postoperative day 15, three mice from each treatment group were randomly selected and sacrificed for tissue collection. Brain tissue was rapidly extracted and fixed in 4% paraformaldehyde at 4°C for 24–48 hours. After fixation, the brain tissue was sequentially dehydrated with graded ethanol, cleared with xylene, and embedded in paraffin. Paraffin sections with a thickness of approximately 5 μm were continuously cut using a rotary microtome and attached to glass slides for later use.

[0167] 1.1 HE staining

[0168] Paraffin sections were dewaxed with xylene and rehydrated with graded ethanol, then stained with hematoxylin and eosin (HE) according to standard methods. After staining, the sections were dehydrated, cleared, and mounted. HE-stained sections were observed and imaged using a pathological microscope.

[0169] 1.2 Ki67 Immunohistochemical Staining

[0170] Paraffin sections used for Ki67 detection were dewaxed, hydrated, and then placed in citrate buffer for antigen retrieval. After blocking, the sections were incubated with Ki67 primary antibody (Anti-Ki67, ab16667, Abcam). Subsequently, goat anti-rabbit IgG secondary antibody (PV-6001, ZSGB-BIO) was added for incubation, and staining was performed using the DAB staining system. After counterstaining, dehydration, clearing, and mounting, the sections were subjected to microscopic imaging. The number of Ki67-positive cells was quantitatively analyzed using ImageJ software.

[0171] 1.3 TUNEL staining

[0172] Apoptosis was detected using the TMR (red) TUNEL apoptosis detection kit (Servicebio, G1502), and the specific procedures were performed according to the kit instructions. Paraffin sections were dewaxed, hydrated, and treated with protease before incubation with TUNEL reaction solution, and the cell nuclei were counterstained with DAPI. Since tumor cells express GFP signals, TUNEL-positive cells with overlapping GFP fluorescence signals in the TUNEL staining were identified as apoptotic tumor cells. The sections were observed and imaged using a fluorescence microscope, and the TUNEL⁺ / GFP⁺ cell count was quantitatively analyzed using ImageJ software.

[0173] To evaluate the therapeutic effect of intraoperative HPAC-siNTRK2@Gel in preventing postoperative GBM recurrence, we established an orthotopic Luci⁺-GL261 glioma model and performed surgical resection under MRI guidance. GL261-luc cells were stereotactically seeded into the right frontal cortex of C57BL / 6 mice. On day 8 post-inoculation, tumor-bearing animals were randomly divided into 6 groups. Preoperative T1-weighted FLAIR MRI scans were performed to locate the tumor and guide precise debulking resection; subsequently, a standardized resection cavity was formed and immediately filled with the appropriate formulation. Figure 10 A). Intraoperative imaging showed that HPAC-siNTRK2@Gel injection could rapidly conform to the irregular shape of the excision cavity within minutes and form close contact with the cavity wall surface. Figure 10 B).

[0174] Histological analysis with H&E staining on postoperative day 15 showed that intraoperative administration of HPAC-siNTRK2@Gel significantly inhibited tumor recurrence compared to all control treatments. Control groups included PBS, HPAC-siNTRK2 solution not encapsulated in hydrogel, hydrogel alone, negative control siRNA-loaded hydrogel, and non-targeted siNTRK2 delivery. Figure 10 C). Consistent with this, Kaplan–Meier survival analysis showed that the HPAC-siNTRK2@Gel group achieved a significant survival benefit, with a median survival of 67 days, and more than 30% of mice were still alive by day 90. Figure 10 D).

[0175] To further characterize the durability of tumor control and systemic tolerability in the GL261 model, we performed longitudinal bioluminescence imaging (BLI) and weight monitoring during postoperative follow-up. BLI results showed that the HPAC-siNTRK2@Gel treatment group exhibited sustained tumor growth inhibition at multiple postoperative time points. Figure 11A, B). Meanwhile, mice receiving HPAC-siNTRK2@Gel experienced only a transient decrease in postoperative body weight, which subsequently stabilized and returned to near baseline levels. Figure 11 C).

[0176] Longitudinal MRI analysis further confirmed the postoperative tumor control effect from a spatial perspective. The tumor volume was comparable across groups before surgery and immediately after resection; however, significant tumor recurrence was observed in control animals on postoperative days 10 and 25. In contrast, the HPAC-siNTRK2@Gel treatment group showed very little tumor regrowth within and adjacent to the resection cavity. Figure 10 E, F). TUNEL staining of GFP-labeled GL261 tumors on postoperative day 15 showed extensive apoptosis in peri-excisional tumor cells after HPAC-siNTRK2@Gel treatment; compared with all control groups, TUNEL signaling colocalization in GFP⁺ tumor cells was significantly increased (E, F). Figure 10 G, H). Consistently, Ki67 immunohistochemical analysis showed that the proliferative activity of recurrent tumor regions was significantly reduced in the HPAC-siNTRK2@Gel group ( Figure 10 I, J).

[0177] To assess whether the therapeutic benefit of intraoperative HPAC-siNTRK2@Gel could be extended to more clinically relevant and significantly heterogeneous glioma backgrounds, we further validated its efficacy in various orthotopic patient-derived glioma stem cell (GSC) xenograft models. These models covered different molecular subtypes, including mesenchymal (G0 147), anterior neural (GSC2), classic (G0 927), and mixed (GSC 1203) GBM (GSC 147). Figure 11 G). In all GSC models, intraoperative HPAC-siNTRK2@Gel administration after surgical resection consistently inhibited postoperative tumor progression (G). Figure 10 K, M; Figure 11 D, F). Accordingly, Kaplan–Meier survival analysis showed that, compared with the surgery-only control group, the HPAC-siNTRK2@Gel treatment group achieved a significant and reproducible prolongation of overall survival in all patient-derived GSC models (D, F). Figure 10 L; Figure 11 E).

[0178] In summary, these results indicate that intraoperative delivery of HPAC-siNTRK2@Gel can stably inhibit postoperative tumor recurrence, promote tumor cell apoptosis, reduce proliferative activity, and significantly prolong survival in syngeneic and multi-patient orthotopic glioma models, supporting its broad applicability as a postoperative treatment strategy for heterogeneous GBM.

[0179] Example 5: Intraoperative HPAC-siNTRK2@Gel treatment for disrupting tumor-neuronal synaptic connections

[0180] 1. Preparation of mouse brain slices for Patch-clamp and Patch-seq recording

[0181] On postoperative day 12, acute brain sections were prepared from the orthotopic glioma mouse model for subsequent Patch-clamp electrophysiological recording and Patch-seq experiments. This experiment included a GFP-labeled GL261 syngeneic transplantation model and a tdTomato-expressing patient-derived glioblastoma stem cell (GSC) xenograft model. Mice were rapidly euthanized by decapitation after isoflurane inhalation anesthesia, and brain tissue was immediately removed and placed in pre-cooled N-methyl-D-glucosamine (NMDG) sectioning solution continuously aerated with 95% O2 and 5% CO2, with the solution temperature maintained at 0–4°C. The NMDG sectioning solution composition was as follows (unit: mM): NMDG 93, HCl 93, KCl 2.5, NaH2PO4 1.2, NaHCO3 30, HEPES 20, glucose 25, sodium ascorbate 5, thiourea 2, sodium pyruvate 3, MgSO4 10, CaCl2 0.5, with the pH adjusted to 7.35. Subsequently, the brain tissue was sectioned into coronal sections approximately 300 μm thick using a vibratory microtome (Leica VT1200S). Immediately after section preparation, the sections were transferred to NMDG sectioning solution at 37.0 ± 0.5 °C with continuous oxygenation for 10–15 min to improve neuronal survival. After resuscitation, the brain sections were transferred to artificial cerebrospinal fluid (ACSF) for equilibration for 1 h before electrophysiological recording. The ACSF solution composition (unit: mM) was as follows: NaCl 12.5, KCl 2.5, NaH2PO4 1.2.5, NaHCO3 2.5, MgCl2 1, glucose 2.5, CaCl2 2, pH adjusted to 7.4, and 95% O2 and 5% CO2 were continuously introduced throughout the equilibration process.

[0182] 2. Whole-cell patch-clamp recording and single-cell sample collection

[0183] Electrophysiological recordings were performed on acute brain slices using whole-cell patch-clamp techniques. To ensure RNA integrity, all lab surfaces and equipment were cleaned sequentially with ddH2O, 75% ethanol, DNA-OFF, and RNase Zap before the experiment. All solutions that came into contact with the RNA samples were prepared under strictly RNase-free conditions.

[0184] Before recording began, the recording chamber was continuously perfused with artificial cerebrospinal fluid (ACSF), and the perfusion fluid was replaced at least 5 minutes before placing the brain slices. The brain slices were placed in the recording chamber, and individual tumor cells were located and recorded under a 40× fluorescence microscope. Tumor cells were identified based on their fluorescent labels, including GFP-labeled GL261 cells in syngeneic transplantation models and patient-derived GSCs expressing tdTomato in xenograft models. Recording was performed using glass microelectrodes with a resistance of 5–8 MΩ. The microelectrode fluid composition was as follows (unit: mM): gluconate 111, KCl 4, HEPES 10, EGTA 0.2, Mg-ATP 4, Na-GTP 0.3, sodium creatine phosphate 5, biotin 13, and a recombinant RNase inhibitor (Takara, 2313A) was added to a final concentration of 1 U / μL. Electrophysiological signals were acquired using a Quadro EPC10 amplifier (HEKA) and controlled and recorded using Patchmaster software (version 2.65). Once a high-resistance seal (≥1 GΩ) is formed, the membrane is ruptured using a slight negative pressure to enter whole-cell mode. Spontaneous excitatory postsynaptic currents (sEPSCs) are recorded under −70 mV membrane potential clamping. Electrophysiological data are analyzed using Mini Analysis software (Version 6.0.3) and Clampfit software (Version 10.4.0.36), respectively.

[0185] After completing electrophysiological recordings, single-cell contents were collected from GL261 brain slices used for patch-seq analysis. Intracellular material was drawn into the recording electrode using negative pressure, and the electrode contents were immediately injected into a 0.2 mL PCR tube pre-filled with 4 μL of fresh lysis buffer. The lysis buffer consisted of 1 U / μL of recombinant RNase inhibitor (Takara, 2313A), 0.33% Triton X-100 (Sigma, T8787), 0.5 mM dNTPs (Thermo Scientific, R1121), and 2.5 μM oligo-dT primers, for subsequent single-cell transcriptome analysis.

[0186] 3. Recording of Schaffer collateral stimulation and evoked synaptic currents in the hippocampus of the GL261 model

[0187] GFP-labeled GL261 cells were stereotactically implanted into the CA1 region of the mouse hippocampus. Acute hippocampal brain sections were prepared as described above for subsequent electrophysiological recording. The sections were placed in the recording chamber and continuously perfused with oxygenated ACSF at 28–30°C. GFP-positive GL261 cells located in the CA1 region were localized under a fluorescence microscope, and whole-cell patch-clamp recording was performed. Synaptic responses were induced by electrical stimulation of Schaffer collateral afferent fibers. Stimulation was performed using concentric bipolar electrodes (125 μm in diameter, FHC) placed within the radiative layer, approximately 100–200 μm from the GL261 cells being recorded. The stimulation signal was triggered by the HEKA system, output via an isolator (AMPI ISO-flex), and the stimulation frequency was set to 0.1 Hz.

[0188] In the paired stimulation experiment, two consecutive electrical stimulations were applied with a 1-second interval. During recording, the GL261 cell voltage was clamped at −70 mV, and the induced excitatory postsynaptic currents (EPSCs) were recorded. Series resistance was continuously monitored throughout the experiment; records exceeding 30 MΩ or showing significant changes during recording were discarded. Electrophysiological signals were amplified using a Quadro EPC10 amplifier (HEKA), digitally acquired at a 10 kHz sampling frequency, and recorded using Patchmaster software (HEKA, version 2.6). Representative current trajectories used for image visualization were de-artifacted during mapping to improve signal clarity.

[0189] 4. Patch-seq single-cell cDNA amplification and library construction

[0190] Single-cell cDNA synthesis and amplification were performed according to the Smart-seq2 workflow. After electrophysiological recording and collection of single-cell contents, single-cell RNA was extracted, and full-length cDNA was synthesized and amplified. The amplified cDNA products were quantified using the Qubit high-sensitivity DNA detection kit (Thermo Fisher, Q32851), and the length distribution of cDNA fragments was detected using the Agilent Bioanalyzer 2100 system (Agilent). CDNA samples meeting quality requirements were used for library construction. Library preparation was performed using the Nextera XT DNA Library Preparation Kit (Illumina, FC-131-1096) and Nextera XT Index Kit V2 Sets A (Illumina, FC-131-2001), following the kit instructions. The constructed libraries were mixed and then sequenced. Sequencing was performed on the Illumina NovaSeq 6000 platform using 150bp paired-end sequencing. The obtained sequencing data were used for subsequent single-cell transcriptome analysis.

[0191] 5. Patch-seq single-cell RNA sequencing data processing and analysis

[0192] Quality control and data processing were performed on the raw single-cell RNA sequencing data obtained from patch-seq. First, FastQC software was used for preliminary quality assessment of the raw sequencing data, obtaining information on sequence quality distribution, base composition, GC content, and adapter contamination. Then, Trimmomatic software (version 0.39) was used to cut the raw sequences, removing adapter sequences and low-quality bases to obtain high-quality, clean data. The processed sequencing data was aligned to the mouse reference genome GRCm38 (mm10) using STAR software (version 2.7.10a). Quantitative analysis of gene expression levels was performed using RSEM software (version 1.3.3). After obtaining the gene expression matrix for each single cell, differential expression analysis was performed on samples under different treatment conditions to screen for genes with significant changes in expression between different treatment groups for subsequent function and mechanism correlation analysis.

[0193] 6. Electrophysiological histological staining and morphological reconstruction

[0194] Immediately after completing electrophysiological recordings and single-cell RNA collection, brain sections underwent subsequent immunohistochemical processing. Brain sections were fixed in 4% paraformaldehyde at 4°C for no more than 24 hours. After fixation, the sections were washed three times with PBS buffer containing 2% Triton X-100. Subsequently, the brain sections were placed in blocking medium containing 10% donkey serum and blocked overnight at 4°C. The next day, the sections were transferred to incubation medium containing 10% donkey serum, with the addition of streptavidin Alexa Fluor™ 594 (Thermo Fisher, S11227, 1:500) and DAPI (Thermo Fisher, 62248, 1:1000), and incubated for two nights in the dark at 4°C. After incubation, the brain sections were mounted with anti-fluorescence quenching mounting medium and covered with glass slides. The sections were imaged using a confocal laser scanning microscope (Nikon AX R, Japan) to record cell morphology and perform subsequent morphological reconstruction analysis.

[0195] 7. Multiplex immunofluorescence staining and synaptic quantification analysis in an orthotopic mouse model of glioma

[0196] Multiplex immunofluorescence staining and quantitative analysis were performed on the tumor-neuronal synaptic structure in various orthotopic glioma mouse models. The models analyzed included a GL261 syngeneic transplantation model and a patient-derived GSC xenograft model representing different molecular subtypes. Brain tissue was harvested at predetermined time points post-surgery: day 15 for the GL261 model and day 28 for the GSC xenograft model. After harvesting, the brain tissue was fixed in 4% paraformaldehyde, routinely embedded in paraffin, and cut into serial sections approximately 5 μm thick.

[0197] Multiplex immunofluorescence staining was performed using the Opal™ 4-Color Manual IHC Kit (Akoya Biosciences), with the specific procedure adjusted for mouse brain tissue based on the manufacturer's instructions. Paraffin sections were dewaxed with xylene and rehydrated with graded ethanol, followed by heat-induced antigen retrieval before each round of antibody incubation. Tyramine signal amplification (TSA) was used for sequential staining, with heat treatment between each staining round to remove antibodies from the previous round. Each staining panel sequentially used antibodies against Nestin (tumor cell marker), NeuN (neuronal marker), PSD95 (postsynaptic density marker), and Synapsin-1 (presynaptic marker), and DAPI was used for counterstaining of cell nuclei. The pairing order of antibodies with Opal fluorescent dyes and the staining procedure were consistent with those used in patient-derived tissue sample analysis to ensure comparability between different experimental systems. Sections that underwent multiplex immunofluorescence staining were imaged using a confocal laser scanning microscope (Nikon AXR) for subsequent spatial analysis and quantification of tumor-neuronal synaptic structures.

[0198] Using an evoked recording paradigm in acute hippocampal sections, stimulation of Schaffer collateral fibers induced time-locked excitatory postsynaptic currents in some GFP⁺GL261 cells, suggesting that these tumor cells can receive functional synaptic input. Figure 12 A–F). Further ultrastructural analysis revealed typical synaptic structural features at the neuron–GL261 cell contact interface. Figure 12 G). Meanwhile, phosphorylated TrkB was enriched in the tumor-brain interface region, suggesting the presence of an active neurotrophic factor signaling pathway in this model. Figure 12 H). The above results collectively confirm the presence of real malignant neuronal glioma synapses within GL261 tumors, and provide an experimental basis for subsequent analysis.

[0199] To investigate how intraoperative HPAC-siNTRK2@Gel treatment affects tumor-neuronal synaptic interactions after surgical resection, we further employed patch-seq technology to simultaneously analyze the electrophysiological and transcriptomic characteristics of single recurrent tumor cells in acute brain slices of a GL261 glioma model. Figure 13 A). To achieve clear identification of tumor cells during whole-cell recording, GL261 cells were pre-labeled with GFP; simultaneously, biotin perfusion was used to link electrophysiological measurements with single-cell RNA sequencing of the same tumor cells. Figure 13 B, C).

[0200] We performed whole-cell patch-clamp recordings on acute brain sections from 12 tumor-bearing mice. Six mice received PBS control treatment, while the other six received intraoperative HPAC-siNTRK2@Gel treatment after tumor resection. A total of 71 GFP⁺ GL261 tumor cells were successfully clamped from these records, of which 39 yielded high-quality RNA suitable for subsequent single-cell RNA sequencing, enabling integrated electrophysiological and transcriptomic analysis at the single-cell level. Figure 13 D, E). Functionally, tumor cells in the PBS-treated group exhibited frequent spontaneous excitatory postsynaptic currents (sEPSCs), suggesting that they could obtain strong excitatory synaptic input from peripheral neurons. Figure 13 D, E). In contrast, the amplitude and frequency of sEPSCs in GL261 cells from mice treated with HPAC-siNTRK2@Gel were significantly reduced, indicating that tumor-neuron synaptic coupling was significantly weakened after treatment. Figure 13 G, H).

[0201] Consistent with electrophysiological results, transcriptomic analysis showed that in the PBS control group, tumor cells with higher frequencies and amplitudes of sEPSCs also expressed higher levels of tumor proliferation-related genes (such as MKI67, ATRX, EGFR, and BRAF), accompanied by upregulation of synapse-related genes, including NTRK2, TTYH2, CACNA1E, CNTN1, HOMER1, CACNA1D, SHANK1, and GRIN2B. Notably, HPAC-siNTRK2@Gel treatment significantly downregulated NTRK2 expression, consistent with the overall decreasing trend of synapse-related gene expression in these tumor cells. Figure 13 D, F). To further link electrophysiological phenotypes with transcriptional state, we constructed a "synapse-related gene expression score" based on the co-expression of synaptic genes in the Patch-seq dataset. In the PBS-treated group, both the amplitude and frequency of sEPSC were positively correlated with this score, suggesting that GL261 cells receiving stronger excitatory synaptic input have more significant synapse-related transcriptional features. Conversely, in the HPAC-siNTRK2@Gel-treated group, this coupling relationship between "synaptic gene expression and functional synaptic activity" was significantly weakened, consistent with the characteristics of malignant tumors – disruption of neuronal synaptic programs (D, F). Figure 13 I, J).

[0202] To assess whether the aforementioned synaptic effects could be generalized beyond the GL261 model, we further conducted electrophysiological recordings in patient-derived GBM xenograft models representing different molecular subtypes. In mesenchymal G0 147 tumors, HPAC-siNTRK2@Gel treatment significantly reduced sEPSC frequency while increasing EPSC amplitude. Figure 13 K, L). In the anterior neural GSC2 xenograft model, both the frequency and amplitude of sEPSCs decreased significantly after treatment (K, L). Figure 13 M, N). In classic G0 927 tumors, neither the frequency nor the amplitude of sEPSC was significantly altered (M, N). Figure 14 A, B). In mixed subtype GSC1203 tumors, the frequency of sEPSC decreased, while the amplitude of EPSC remained largely unchanged. Figure 14 C, D).

[0203] Finally, multiplex immunofluorescence results showed that in the PBS-treated group, Nestin⁺ glioma cells recurring at the resection margin were extensively intertwined with neuronal processes in space, and high-density, adjacent distribution of punctate signals of presynaptic markers (Synapsin-1) and postsynaptic markers (PSD95) was observed at the tumor-brain boundary. After intraoperative HPAC-siNTRK2@Gel treatment, the tight spatial association between Nestin⁺ tumor cells and synaptic marker-positive neural structures was significantly weakened, and the interweaving of tumor and neuronal structures at the interface was reduced. Figure 13 O, P; Figure 14 E–G).

[0204] In summary, multidimensional evidence from electrophysiology, transcriptomics, and spatial distribution consistently demonstrates that intraoperative HPAC-siNTRK2@Gel treatment can effectively disrupt the functional synaptic coupling between tumor cells and neurons after surgical resection.

[0205] Example 6: HPAC-siNTRK2@Gel synergistically inhibits tumor invasion and prolongs survival in a patient-derived GBM xenograft model with radiotherapy and chemotherapy.

[0206] 1. Combined chemotherapy and radiotherapy treatment regimen

[0207] To evaluate the therapeutic efficacy of HPAC-siNTRK2@Gel combined with standardized radiotherapy and chemotherapy, experiments were conducted in patient-derived glioblastoma xenograft models, including the G0 147 mesenchymal GBM model and the G0 927 classical GBM model. After successful modeling and tumor resection, mice were randomly assigned to the following treatment groups: a surgery-only control group, the HPAC-siNTRK2@Gel group, the temozolomide (TMZ) group, the radiotherapy group, and the corresponding combination therapy group. TMZ was administered intraperitoneally at a dose of 10 mg / kg / day, following a cyclical regimen of 5 days of administration followed by 2 days of rest, for a total treatment duration of 21 days. Radiotherapy was performed using fractionated local cranial irradiation under isoflurane anesthesia, with a daily dose of 3 Gy for 21 consecutive days. The sham irradiation group served as a control, receiving only the same anesthesia and fixation procedures without radiation. In the combination therapy group, the dosage and timing of TMZ and radiotherapy were consistent with those in the monotherapy group, and HPAC-siNTRK2@Gel was administered intratumorally immediately after tumor resection.

[0208] During treatment, bioluminescence imaging was used to monitor tumor progression longitudinally, with the same imaging parameters used in each group to ensure comparability. Mice used for mechanistic studies and histological analysis were sacrificed on day 14 post-surgery, and brain tissue was harvested for pathological and immunofluorescence staining. The focus was on assessing the extent of tumor infiltration, cell proliferation and apoptosis, as well as changes in the expression of synaptic markers at the tumor-brain interface.

[0209] To evaluate whether intraoperative HPAC-siNTRK2@Gel is compatible with standard treatment regimens and can enhance their antitumor effects, we systematically examined its therapeutic performance in an orthotopic patient-derived GBM xenograft model after surgical resection. Patient-derived GBM stem-like cells (GSCs) labeled with luciferase and tdTomato were stereotactically seeded into the brains of NSG mice, followed by tumor reduction surgery. Postoperatively, mice were treated with HPAC-siNTRK2@Gel, radiotherapy, temozolomide (TMZ), or a combination thereof. Figure 16 A). In this patient-derived GSC model derived from G0 147, HPAC-siNTRK2@Gel significantly delayed tumor recurrence and prolonged survival; efficacy was further enhanced when combined with TMZ or radiotherapy, with the triple regimen showing the most significant benefit. Figure 16 B–D).

[0210] It is noteworthy that while HPAC-siNTRK2@Gel combined with TMZ was superior to TMZ monotherapy, it did not show any additional benefit compared to HPAC-siNTRK2@Gel monotherapy, suggesting that the synergistic effect between the two may be limited. In contrast, HPAC-siNTRK2@Gel combined with radiotherapy achieved better tumor control than either treatment alone, suggesting a possible synergistic effect. Given previous studies suggesting that radiotherapy may increase tumor invasiveness, we further compared the tumor infiltration characteristics of each group. Compared to the surgery-only control group, radiotherapy significantly increased the number and total area of ​​detached tumor cell clusters at the tumor-brain interface. It is worth emphasizing that the combined administration of HPAC-siNTRK2@Gel effectively counteracted the pro-invasive effect of radiotherapy. Figure 16 E,F).

[0211] Immunofluorescence analysis further showed that radiotherapy combined with HPAC-siNTRK2@Gel reduced tumor cell proliferation (Ki67⁺), increased apoptosis (cleaved caspase-3⁺), and significantly expanded the spatial overlap between the two, suggesting that this combined regimen may be more inclined to eliminate tumor cells in the active proliferation cycle. Figure 16 G, H). Analysis of synaptic markers showed that, compared with radiotherapy alone, combination therapy significantly reduced the fluorescence intensity of Synapsin-1 and PSD95 at the tumor-brain interface (G, H). Figure 16 I, J).

[0212] To assess whether the synergistic effect of HPAC-siNTRK2@Gel with radiotherapy could be generalized to different GBM subtypes, we further validated the efficacy of HPAC-siNTRK2@Gel in another independent orthotopic xenograft model. This model originated from GSC patients G0 927, representing classic GBM. In this model, HPAC-siNTRK2@Gel combined with radiotherapy also significantly reduced tumor burden, and was superior to radiotherapy alone. Figure 17 ).

[0213] In summary, HPAC-siNTRK2@Gel can not only delay GBM recurrence and prolong survival as a single drug, but also enhance the anti-tumor effect in synergistic effect with standard postoperative radiotherapy and chemotherapy regimens.

[0214] Our data also suggest that local disruption of malignant neuronal-tumor synapses not only inhibits tumor growth itself but also functionally counteracts radiotherapy-induced enhanced invasiveness. Previous studies have shown that neuronal activity and synaptic input can promote tumor microtubule formation, thereby supporting tumor invasion and conferring radiotherapy tolerance. By disrupting synaptic-driven processes at the tumor-brain interface, intraoperative TrkB silencing may inhibit the formation of a treatment-resistant tumor network interconnected by tumor microtubules, thereby restoring the cytotoxic effects of radiotherapy.

Claims

1. The use of siRNA-loaded hydrogels in the preparation of medicaments for the treatment and / or adjuvant treatment of GBM, characterized in that, The hydrogel is obtained by mixing OHA, CMC and HPAC-siNTRK2 in a certain proportion. The formulation of the hydrogel is 3% OHA + 5% CMC + 3% HPAC-siNTRK2. The preparation process of HPAC-siNTRK2 is as follows: S1. Synthesis of HPAA-PEG-MAL: HPAA undergoes an amidation reaction with Mal-amido-PEG1-C2-NHS to obtain HPAA-PEG-MAL; S2. Synthesis of HPAC: HPAA-PEG-MAL and HS-PEG-CRGD undergo a Michell addition reaction to yield HPAC; S3, HPAC-siNTRK2 complex: The HPAC-siNTRK2 complex is obtained by mixing and incubating the HPAC and siNTRK2 described in S2; wherein the mass ratio of HPAC to siNTRK2 is 20~80:1; The sequence of siNTRK2 described in S3 is shown in SEQ ID NO: 1 and SEQ ID NO:

2.

2. The application according to claim 1, characterized in that, The siRNA-loaded hydrogel and / or radiotherapy, and / or TMZ in combination.

3. The application according to claim 1, characterized in that, The GBM is mesenchymal type G0 147 and classical type G0927.

4. The application according to claim 2, characterized in that, The hydrogel loaded with siRNA, when used in combination with / or radiotherapy and / or TMZ, has at least one of the following effects compared to any single treatment: (1) Prolong the survival of GBM after surgery, or; (2) Delaying postoperative GBM tumor recurrence, or; (3) Promotes postoperative apoptosis of GBM tumor cells, or; (4) Reduce postoperative GBM tumor cell proliferation, or; (5) Reduce tumor invasion caused by radiotherapy, or; (6) Disrupts the functional synaptic coupling between tumor cells and neurons after surgery.

5. The application according to claim 4, characterized in that, The promotion of postoperative GBM tumor cell apoptosis refers to the promotion of tumor cell cleaved caspase-3. + Apoptosis.

6. The application according to claim 4, characterized in that, The reduction of postoperative GBM tumor cell proliferation refers to reducing the Ki67 of tumor cells. + proliferation.

7. The application according to claim 4, characterized in that, The disruption of functional synaptic coupling between tumor cells and neurons after surgery refers to reducing the fluorescence intensity of Synapsin-1 and PSD95 at the tumor-brain interface.

Images