Radiation-activated astrocytes promote tumor progression through CXCL12 secretion and induce resistance to anti-PD-1 immunotherapy

Introduction

Glioblastoma (GBM), the most lethal primary brain tumor with a median survival of only 15 months despite aggressive treatment, remains a formidable clinical challenge due to its invasive growth pattern and remarkable intratumoral heterogeneity (1). The current standard-of-care, which involves maximal safe resection followed by concurrent radiotherapy (60 Gy in 30 fractions) and temozolomide (Stupp protocol), results in only modest improvements, and 5-year survival rates remain persistently below 10% (2). For high-grade gliomas (HGGs), radiotherapy serves as an irreplaceable cornerstone of multidisciplinary therapy (3). However, 80–90% of recurrences occur within the 2-cm margin of the original tumor bed, recognizing as a region previously exposed to sublethal radiation doses (4). At present, glioma research predominantly concentrates on primary gliomas, particularly those that are newly diagnosed, while less attention is given to recurrent gliomas. If gliomas were effectively eradicated at their origin, recurrence would not be a concern. Alternatively, in cases where gliomas are highly aggressive, the primary tumor takes precedence, leaving little time to address the possibility of recurrence. As a result, unlike primary gliomas, there is no established standard treatment for recurrent gliomas (5). However, nearly all gliomas may eventually recur, and recurrent gliomas are typically more lethal and exhibit reduced sensitivity to treatments compared with primary tumors, making them a critical concern (6). This progression is partly due to selective pressures exerted by the microenvironment, such as the effects of treatment, immune responses, and evolutionary forces, that further contribute to GBM heterogeneity. Additionally, radiotherapy serves as a selective pressure, potentially enhancing glioma cell survival (7). Rather than regarding GBMs as homogeneous, static tumors with fixed genetic traits, gliomas are progressively being recognized as dynamic entities that evolve and adapt over time (8). Hence, there is no standard treatment for relapsed gliomas, and these tumors may exhibit distinct therapeutic responses compared with primary gliomas.

Cancer arises from mutations accumulated in cancer cells; however, disease progression and response to treatment are remarkably influenced by non-mutant cells in the tumor microenvironment (TME). These cells modulate cancer metastasis through the synthesis and remodeling of the extracellular matrix (ECM), as well as the production of growth factors. They also influence angiogenesis, tumor mechanics, drug uptake, and therapeutic responses (9). The role of stromal cells in the TME is increasingly recognized as a critical factor in tumor development, rather than merely acting as passive bystanders (10). There are several types of tumor stromal cells, including tumor-associated mesenchymal stem cells (TA-MSCs), cancer-associated fibroblasts (CAFs), tumor-associated macrophages (TAMs), and tumor endothelial cells (TECs). TA-MSCs exert carcinogenic effects through a variety of mechanisms, including promoting tumor growth, metastasis, tumor cell dryness, chemotherapy resistance, and immunosuppression (11). TA-MSCs release inflammatory cytokines and chemokines, including interleukin (IL)-6, IL-8, vascular endothelial growth factor (VEGF), and CXCL12, which promote tumor growth and enhance its carcinogenic properties (12). A recent study has demonstrated that several carcinogenic and treatment-resistant features of the TME can be attributed to fibroblast activity (13). CAFs contribute to the malignant progression of breast cancer by activating the mTOR/FIP200/ATG13-induced PDGFR-β/Cav1 autophagy pathway (14). COL1A1+ TECs undergo EndoMT through the upregulation of CEBPB, driving tumor invasiveness. Additionally, COL1A1+ TECs interact with malignant cells via the ANGPTL4-SDC4 axis, enhancing invasion and migration (15).

Astrocytes, the most abundant type of glial cells in the central nervous system (CNS), exhibit complex morphology (16,17). These cells play a crucial role in regulating CNS inflammation and neurodegeneration through multiple mechanisms, including neurotoxicity, modulation of microglial activities, recruitment of inflammatory cells into the CNS, and even via their metabolic cascade (18). Tumor-associated astrocytes (TAAs) have recently been found to be involved in the formation of the TME in both primary and secondary brain tumors (19,20). It was demonstrated that cholesterol from astrocytes is the key to the survival of glioma cells, and targeting ABCA1-mediated cholesterol efflux from astrocytes can prevent tumor progression (18). In addition to playing an important role in healthy tissue, astrocytes exhibit a primitive evolutionary response known as astrocyte reactivity in response to CNS injuries (21). However, the role of reactive astrocytes in the pathogenicity of GBM has been poorly understood. Radiotherapy induces damage to astrocytes, triggering a reactive astrocyte response. The impact of this reaction on recurrent glioma warrants further assessment.

Experiments have demonstrated that astrocytes significantly increased CXCL12 secretion. Elevated CXCL12 acts on irradiated glioma cells, promoting their survival and contributing to radio-resistance through a mechanism involving CXCL12-induced inhibition of autophagy in post-radiotherapy glioma cells. Furthermore, in our mouse model of radiotherapy-recurrent glioma, significantly increased infiltration of CD4⁺ T cells, CD8⁺ T cells, natural killer (NK) cells, and macrophages was identified compared with primary glioma, which was associated with a worse prognosis. These recurrent tumors exhibited a relatively poor response to programmed cell death protein 1 (PD-1) antibody therapy. This finding may highlight the pivotal significance of implementing immunotherapy for primary gliomas, while also suggesting a diminished role of immunotherapy in the management of recurrent gliomas. This finding may guide the importance of immunotherapy for radiotherapy-recurrent gliomas. We present this article in accordance with the ARRIVE and MDAR reporting checklists (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2652/rc).

Methods

Cells and cell culture conditions

LN229, U87MG (ATCC version, Cat. No. HTB-14), GL261, and normal human astrocytes (NHAs) were obtained from the Chinese Academy of Science and were maintained at the Key Laboratory of Neurosurgery Affiliated to Second Hospital of Lanzhou University (Lanzhou, China). The cells were maintained in a complete Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (Cat. No. KGL3006-50; KEYGEN, Nanjing, China), 100 U/mL penicillin, and 100 µg/mL streptomycin. Culture conditions were set as follows: 37 ℃, 5% CO₂, and constant humidity. The culture medium used was DMEM with high glucose (4.5 g/L D-glucose), L-glutamine, and 0.110 g/L sodium pyruvate, without penicillin-streptomycin. All cell lines used were authenticated by short tandem repeat (STR) profiling, and the STR profile of U87MG matched the ATCC reference database.

Animals

In this study, C57BL/6 and nude mice (BALB/c nu/nu) were obtained from the Lanzhou Veterinary Research Institute (LVRI), Chinese Academy of Agricultural Sciences (Lanzhou, China). Mice were housed in the laboratory animal facilities of the Neurosurgery Laboratory at the Second Hospital of Lanzhou University. All mice were male, 4-week-old, and allowed a 2-week acclimatization period prior to experimentation. They were housed five per cage under a standard 12-hour light/dark cycle and had ad libitum access to food and water. For in vitro experiments, mice were killed by cervical dislocation. All experimental procedures were approved by the Institutional Animal Care and Use Ethics Committee of Lanzhou University Second Hospital (No. D2023-034), in compliance with the ethical guidelines set by Lanzhou University Second Hospital for the care and use of animals, and all the experiments were conducted in accordance with the Laboratory Animal Guidelines for Ethical Review of Animal Welfare. The animals were randomly assigned to each experimental group using the computer-generated random number method. A protocol was prepared before the study without registration. The cage positions for all mice and the sequence of treatments were controlled using the computer-generated random number method.

Irradiation of cells and mice

Irradiation of both cells and animals was conducted at the Institute of Modern Physics, Chinese Academy of Sciences using the X-RAD 225 biological irradiator system (Precision X-Ray Inc.). For mouse experiments, cranial irradiation was administered at total doses of 10 or 15 Gy, depending on the specific experimental conditions. Multiple irradiations were administered, with each session delivering a dose of 2 Gy, conducted at one-day intervals until the target dose was achieved. During the irradiation of mice, they were placed in a lead container with only the targeted area exposed for irradiation. In contrast, all cells were exposed to a uniform radiation dose of 4 Gy, delivered in fractions of 2 Gy per session. Glioma cells were processed shortly after completing irradiation, while the medium from RT-astrocytes was harvested 24 hours post-irradiation of the astrocytes.

Immunohistochemical staining

The tumor samples were fixed with formaldehyde for 1 week, dehydrated using a fully enclosed tissue processor (Leica, ASP300), and embedded in paraffin. The slices were cut into 4 µm slices on an automatic microtome and then placed in an oven at 65 ℃ for 6 h. The sections were then dewaxed through xylene, absolute ethanol, and ethanol dehydration. The sections were then immersed in 0.01 M citric acid repair solution (pH 6.0) and heated in a microwave oven at high power for 15 min for antigen repair. Allow to cool to room temperature and wash with phosphate-buffered saline (PBS). The cells were blocked with 5% goat serum for 20 min at room temperature. The sections were thereafter incubated overnight at 4 ℃ with the primary antibody, followed by a 30-min incubation at 37 ℃ with BOC biotinylated secondary antibody. The primary antibodies used included 8-OHdG (Abcam, Cambridge, UK; ab81299, 1:100), γ-H2AX (Abcam, Ab62623, 1:100), CD4 (Abcam, Ab183685, 1:200), CD8 (Abcam, Ab217344, 1:200), CD56 (Abcam, Ab220360, 1:200), PD1 (Abcam, Ab52587, 1:200), CD68 (Servicebio, Wuhan, China; GB113109. 1:200), CXCL12 (Proteintech, 17402-1-AP, 1:200), Ki67 (Huabio, Woburn, MA, USA; HA721115, 1:400), and Caspase3 (Servicebio, GB11532, 1:500). After washing the sections three times with PBS, 3,3’-diaminobenzidine (DAB) staining reagent (concentrated DAB kit, K135925C, Zhongshan Jinqiao Biotechnology Co., Ltd., Beijing, China) was applied, and color development was monitored under an optical microscope. Excessive DAB stain was removed by rinsing with distilled water (typically after 2 min of staining). After washing the slides, 100 µL hematoxylin was added and incubated for 3 min (the time could be adjusted appropriately according to the color development). After washing the slides, they were dehydrated and cleared with different gradients of ethanol, absolute ethanol and xylene, followed by neutral glue to seal the sections. Sections were initially observed at 100× magnification using an optical microscope (DMI8, Leica, Berlin, Germany) to select appropriate fields. Finally, images from three fields at 400× magnification were collected. The light density and area of the images were measured and recorded according to the specified conditions. The average light density for each image was calculated, or the number of positive cells in each image was counted.

Immunofluorescence assay

Cells were cultured in 24-well plates (3×10⁴ cells/well), each well containing a glass coverslip with a 14 mm diameter at the bottom. After the cells were completely attached to the wall, then washed with precold PBS and fixed with 4% paraformaldehyde (PFA) for 30 min. The cells were then incubated with 0.1% Triton X-100 for 5 min at RT then washed with PBS for 10 min. Cells were blocked with 10% goat serum at room temperature for 1 h, followed by incubation with primary antibodies against the target protein at 4 ℃ overnight. The next day, the cells were incubated with fluorescent secondary antibodies for 2 h at room temperature in the dark. The nuclei were counterstained with 4’,6-diamidino-2-phenylindole (DAPI), and the slides were examined under a fluorescence microscope (Leica DMI8) in a darkroom setting.

Western blotting

Cells were lysed on an ice-cold surface using radio-immunoprecipitation assay (RIPA) lysis buffer (Beyotime, Shanghai, China) containing protease and phosphatase inhibitors at a 1:100 ratio. The total protein concentration was thereafter determined using a bicinchoninic acid (BCA) kit (Beyotime). The sample quantities for different electrophoresis lanes were adjusted according to the protein concentration. Following that, the proteins underwent separation by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE, Beyotime, Shanghai, China) and were transferred onto a polyvinylidene fluoride (PVDF) membrane (Millipore, Burlington, MA, USA). The PVDF membrane was blocked with 5% skimmed milk, washed with TBST, and incubated with primary antibodies overnight at 4 ℃. The following day, the samples were incubated with HRP-labeled secondary antibodies (Proteintech, Wuhan, China). After washing with tris-buffered saline with Tween 20 (TBST), the enhanced chemiluminescence reagent was applied for exposure.

Enzyme-linked immunosorbent assay (ELISA)

The culture medium was collected and stored at −80 ℃. The CXCL12 concentration in the medium was measured using an ELISA kit (Abcam) following the manufacturer’s instructions.

Flow cytometry analysis

LN229 and U87MG cells in the logarithmic growth phase were harvested by aspirating the supernatant, washing with PBS, and digesting with trypsin. After centrifugation at 250 ×g for 5 min, the supernatant was discarded and the cells were resuspended in an appropriate culture medium. Following cell counting, cells were seeded into 6-well plates at a density of 2×10⁵ cells/well (2 mL/well) and cultured at 37 ℃ with 5% CO₂. After treatment, the supernatant was aspirated and the cells were washed once with PBS. Cells were then digested with trypsin [without ethylenediaminetetraacetic acid (EDTA)] until detachment occurred. The digestion was terminated by adding the collected conditioned medium, followed by centrifugation at 250 ×g for 5 min. The pellet was washed with PBS, transferred to 1.5 mL microcentrifuge tubes, and recentrifuged at 250 ×g for 5 min to obtain cell pellets. For apoptosis detection, cell pellets were resuspended in 500 µL binding buffer. Then, 5 µL of Annexin V was added and gently mixed, followed by the addition of 5 µL of propidium iodide (PI). After 5-min incubation at room temperature in the dark, samples were analyzed by flow cytometry. For reactive oxygen species (ROS) detection, DCFH-DA was diluted 1:1,000 in a serum-free medium to achieve a final concentration of 10 µmol/L. Then, 1 mL of the diluted DCFH-DA solution was added to each tube and incubated at 37 ℃ for 20 min, with gentle mixing every 3–5 min. Cells were centrifuged at 350 ×g for 5 min, thrice washed with serum-free medium, and then resuspended for analysis. For the positive control, after loading with DCFH-DA, Rosup was diluted 1:1,000 in a serum-free medium, and 1 mL of the solution was added to the cells, followed by 30-min incubation at 37 ℃. Cells were washed 3 times with the serum-free medium, centrifuged at 350 ×g for 5 min, resuspended in PBS, and analyzed by flow cytometry.

Cell Counting Kit-8 (CCK-8) assay

RT-LN229 and RT-U87MG cells that had undergone different treatments (varying concentrations of CXCL12, different astrocyte culture media, and with or without the addition of CXCL12 antibodies) were collected by trypsinization after being washed with PBS. After centrifugation at 250 ×g for 5 min, the supernatant was removed and cells were resuspended in an appropriate culture medium to prepare single-cell suspensions. Following cell counting, cells were seeded at a density of 1×10⁵ cells/well and cultured for 24 h after grouping treatments. The supernatant was thereafter carefully aspirated. A 1:10 dilution of CCK-8 reagent was prepared using a serum-free medium, and 110 µL of the diluted CCK-8 working solution was added to each well. The plate was gently shaken several times to ensure proper mixing and incubated for 2 h at 37 ℃ with 5% CO₂. Finally, the absorbance of each well was measured at 450 nm using a microplate reader.

Transmission electron microscopy (TEM)

After grouping the cells, the culture medium was discarded, and the cells were washed three times with PBS. They were then digested with EDTA-free trypsin at 1,000–1,500 rpm for 5 min. Next, 2.5% glutaraldehyde fixative was added to the cell suspension, and fixation was carried out at room temperature for 30 min. Following fixation, post-fixation was carried out using 1% osmium tetroxide to further stabilize the tissue structure. The cells were then centrifuged at 1,500 rpm for 5 min, and the fixative was discarded. PBS was added to the cells, and they were stored at 4 ℃. A stepwise dehydration process was implemented with acetone, employing a well-defined concentration gradient of 30% → 50% → 70% → 80% → 90% → 95% → 100% to ensure the complete removal of water from the samples while minimizing structural damage. For the infiltration step, the dehydrating agent (acetone) and Epon-812 embedding medium were combined in a sequential manner according to the ratios of 3:1, 1:1, and 1:3, respectively. This step-by-step infiltration process facilitated the gradual replacement of acetone with the embedding medium, ensuring optimal penetration and uniform distribution within the tissue. Following infiltration, the samples were embedded using pure Epon-812 embedding medium, which provided a stable and well-defined matrix for subsequent ultrathin sectioning. Ultrathin sections with a thickness ranging from 60 to 90 nm were precisely prepared using an ultramicrotome and then carefully transferred onto copper grids, which served as the support for the sections during the staining and microscopic examination procedures. Staining was performed in two steps. Firstly, the sections were stained with uranyl acetate for 10–15 min. Uranyl acetate is a commonly used heavy metal stain that binds to negatively charged components in the tissue, enhancing the contrast of cellular structures. Subsequently, the sections were stained with lead citrate for 1–2 min at room temperature. Lead citrate also contributes to contrast enhancement by binding to different tissue components, providing a complementary staining effect to uranyl acetate. Finally, the stained sections were subjected to microscopic examination using an appropriate electron microscope, enabling the detailed visualization and analysis of the cellular and subcellular structures within the samples. The detection was performed by Lilai Biotechnology Co., Ltd.

In vivo mouse experiments

The models were established in specific pathogen-free (SPF) laboratories. A subcutaneous tumor model was established using nude mice. LN229 cells in the logarithmic growth phase, pre-irradiated with 4 Gy, were harvested and resuspended in PBS at a concentration of 1×10⁶ cells/100 µL. After disinfecting the dorsal/axillary skin, 0.1 mL of the cell suspension was subcutaneously injected, and equal amounts of either normal astrocytes or irradiated astrocytes were co-injected according to the experimental groups. Mice received a total radiation dose of 10 Gy at 15 days post-inoculation, and they were euthanized on day 22 for analysis. An in vivo intracranial tumor model was established using a stereotactic technique, in which 5 µL of GL261 (4 Gy) cell suspension was slowly injected into the brains of C57BL/6 mice at a rate of 1 µL/min. The injection site was positioned 1 mm posterior to the fontanelle, 2 mm lateral to the median suture, and at a depth of 3.5 mm. Subsequently, an MRI was performed on day 15 (uMR 9.4T, United Imaging Life Science Instrument, Wuhan, China). Mice are anesthetized via inhalation of sevoflurane. On day 6 post-modeling, mice began receiving anti-PD-1 treatment at a dose of 200 µg per mouse every 3 days.

Immune infiltration analysis

The data were sourced from the Chinese Glioma Genome Atlas (CGGA) public database (http://cgga.org.cn). Transcriptome data from the CGGA cohort were obtained. Based on this dataset, we divided the samples into two cohorts—high expression and low expression—using the median value of CXCL12 expression as the cutoff. The infiltration levels of 22 immune cell subsets in tumor tissues were analyzed using the CIBERSORT algorithm.

Lactate dehydrogenase (LDH) experiment

This study utilized mouse CD8⁺ T cells purchased from Procell (Wuhan, Chin) and GL261 cells to conduct an LDH assay, aiming to evaluate the cytotoxic effect of CD8⁺ T cells on GL261 cells and the influence of CXCL12 recombinant protein. First, CD8⁺ T cells were cultured in CD8⁺ T cell complete medium (Procell, CM-M380) and then activated using IL-2 and CD3/CD28 antibodies. Subsequently, they were mixed with GL261 cells at a ratio of 50:1. In terms of experimental system setup, on one hand, a system containing only GL261 cells was constructed as a basal control to reflect the natural LDH release of GL261 cells. On the other hand, a system with a mixture of CD8⁺ T cells and GL261 cells was established, and an additional group with the addition of 50 ng/mL CXCL12 recombinant protein was included to observe the effect of CXCL12. Meanwhile, a system was set up by adding an appropriate amount of 10% Triton X-100 to the GL261-containing mixture 1 hour before detection to achieve a final concentration of 1%. This system served as a maximal release control to calibrate the baseline for LDH release. After co-culturing the relevant cell systems for 48 hours, an LDH kit was used to detect the LDH levels in each system. The kill rate was calculated based on the detection results using the following formula: Kill rate = (LDH value of the experimental group − LDH value of the group containing only GL261 cells)/(LDH value of the maximal release control group − LDH value of the group containing only GL261 cells). Through this calculation method, the kill rate of CD8⁺ T cells on GL261 cells under different conditions could be determined, thereby analyzing the impact of CXCL12 on the cytotoxic function of CD8⁺ T cells.

Statistical analysis

Throughout all experimental phases, both the operators and the analysts were blinded to the specific details of the experimental groupings. The data were analyzed using SPSS 21.0 software (IBM, Armonk, NY, USA), employing either one-way analysis of variance (ANOVA) or an independent samples t-test. Data were expressed as the mean ± standard deviation. Levene’s test was used to examine the homogeneity of variance. Multiple comparisons were performed using the least significant difference (LSD) method. P<0.05 was considered to indicate a statistically significant difference.

Results

Radiotherapy induced astrocytes to release a higher amount of CXCL12

Three mice were randomly assigned to each group. Radiotherapy was administered to the heads of mice, with total radiation doses set at 10 and 15 Gy, respectively, and a daily dose limit of 2 Gy. Immunohistochemistry was used to assess CXCL12 expression level in the cerebral cortex. The results revealed a significantly higher CXCL12 expression level in the cerebral cortices of mice subjected to 10 and 15 Gy radiotherapy compared with those that did not receive radiotherapy. Additionally, immunofluorescence was employed for co-localization analysis of CXCL12 and the astrocyte marker GFAP, confirming an increased CXCL12 expression level in astrocytes following radiotherapy. A similar pattern was found in the hypothalamus region (Figure 1A,1B). At the cellular level, validation was conducted through Western blotting. The findings demonstrated that, in comparison to the normal untreated astrocyte cell line, astrocytes that had undergone radiation treatment (RT-astrocytes) exhibited elevated expression levels of CXCL12 (Figure 1C). Under immunofluorescence microscopy, RT-astrocytes exhibited increased green fluorescence, indicating a higher CXCL12 expression level (Figure 1D). Given that CXCL12 is a secreted protein, ELISA was employed to quantify its concentration in the cell culture supernatant. The results consistently revealed that the amount of CXCL12 released into the culture medium by RT-astrocytes was significantly higher than that by normal astrocytes (Figure 1E). All experimental assays were performed 24 hours post-irradiation of astrocytes.

Figure 1 Radiotherapy induces upregulation of CXCL12 expression level in astrocytes. (A) IHC and IF analyses of CXCL12 expression level in the cerebral cortex of mice following exposure to varying radiation doses (n=3). (B) Quantification of CXCL12 expression level in the hypothalamus of irradiated mice via IHC and IF staining across different radiation doses (n=3). (C) Western blot assessment of CXCL12 protein level in primary astrocytes versus irradiated astrocytes (RT-astrocytes) (n=3). (D) IF visualization of CXCL12 localization in astrocytes and RT-astrocytes (n=3). (E) ELISA of secreted CXCL12 concentrations in culture medium from astrocytes and RT-astrocytes (n=3). Scale bars: 50 µm for IHC images; 20 µm for IF images. *, P<0.05; **, P<0.01 vs. control group. DAPI, 4’,6-diamidino-2-phenylindole; ELISA, enzyme-linked immunosorbent assay; IF, immunofluorescence; IHC, immunohistochemical; RT, radiation‑treated.

The promoting effects of the culture medium from RT-astrocyte on RT-glioma cells

Given the higher CXCL12 concentration in the RT-astrocyte culture medium, its potential effects were further investigated using this medium. Subsequent experiments were carried out on two glioma cell lines, LN229 and U-87MG. Immunofluorescence staining of Ki67 revealed that both the culture medium from normal astrocytes and the RT-astrocyte culture medium led to increased Ki67 expression levels in irradiated glioma cells. The effect was more notable with the RT-astrocyte medium, indicating the enhanced proliferative activity in these cells (Figure 2A). Although irradiated glioma cells exhibited some degree of apoptosis and ROS accumulation, both the normal and RT-astrocyte culture media effectively reduced apoptosis and mitigated ROS buildup in the irradiated glioma cells. Importantly, the RT-astrocyte culture medium demonstrated a superior effect compared with the astrocyte culture medium (Figure 2B,2C). We hypothesize that CXCL12 in the culture medium may activate relevant pathways in glioma cells that are anti-apoptotic and pro-proliferative, keeping the cells in an active proliferative state, which results in reduced apoptosis and a decrease in ROS. Radiotherapy exhibits a certain cytotoxic effect on glioma cells. Following radiation exposure, normal glioma cells underwent radiation-induced damage, marked by the accumulation of biomarkers for radiation-induced damage, namely γ-H2AX and 8-OHdG. Both the astrocyte culture medium and the RT-astrocyte culture medium significantly reduced the levels of these biomarkers in irradiated glioma cells. This finding suggests that these culture media could induce a certain degree of radio-resistance in glioma cells (Figure 2D,2E). Using the CCK-8 assay, the viability of glioma cells was evaluated 24 h post-radiotherapy. The results revealed that both the astrocyte culture medium and the RT-astrocyte culture medium significantly enhanced the survival rate of the irradiated glioma cells. Given that statistically significant differences were already identified at the 24-h time point, the assessment was not extended to 48 and 72 h (Figure 2F). Additionally, by measuring CXCL12 level in the culture media of irradiated glioma cells, glioma cells co-cultured with astrocytes, and glioma cells treated with the RT-Astrocyte culture medium, it was revealed that the irradiated glioma cells also exhibited a notable capacity for CXCL12 release (Figure 2G). In fact, their CXCL12 secretion level was even higher than that of normal astrocytes and comparable to that of RT-astrocytes (Figure 1E).

Figure 2 Culture medium from RT-astrocytes enhances proliferation, inhibits apoptosis, attenuates ROS accumulation, and promotes radio-resistance and survival in irradiated glioma cells (RT-LN229, RT U-87MG). (A) Immunofluorescence assay used for detecting the Ki67 (n=3). (B) Flow cytometry was used to detect apoptosis (n=3). (C) Flow cytometry was used to detect ROS (n=3). (D) Immunofluorescence assay was used for detecting the γ-H2AX (n=3). (E) Immunofluorescence assay was employed for detecting the 8-OHdG (n=3). (F) 24 h CCK-8 assay results (n=3). (G) CXCL12 concentrations in culture medium were quantified using ELISA (n=3). Scale bars: 20 µm for IF images. ●, RT LN229 (RT U-87MG) group; ■, RT LN229 (RT U-87MG) + astrocytes culture medium group; ▲, RT LN229 (RT U-87MG) + RT-astrocytes culture medium group. *, P<0.05; **, P<0.01; ***, P<0.001. ELISA, enzyme-linked immunosorbent assay; ROS, reactive oxygen species; RT, radiation‑treated.

Activation of the CXCL12/CXCR4 pathway and inhibition of autophagy in glioma cells

When irradiated glioma cells were cultured in the culture media derived from astrocytes and RT-astrocytes, a significant elevation in the expression levels of CXCL12/CXCR4 pathway-associated proteins in the glioma cells was identified (Figure 3A). Furthermore, a clear upward trend in the phosphorylation ratios of p-PI3K/PI3K and p-mTOR/mTOR indicated activation of the PI3K/mTOR signaling pathway (Figure 3B). Concurrently, Western blot analysis of autophagy-related proteins revealed that Beclin-1 expression level and LC3 II/LC3 I ratio were reduced, indicating that the culture media from astrocytes and RT-astrocytes could suppress autophagy in glioma cells (Figure 3C). Furthermore, TEM observations confirmed a decline in the number of autophagosomes and autolysosomes (Figure 3D).

Figure 3 RT-astrocyte derived culture medium upregulates the CXCL12/CXCR4 chemokine axis in RT LN229 cells and RT U-87MG cells subjected to radiotherapy, subsequently inhibiting autophagy through PI3K/mTOR pathway. (A) Western blotting confirmed upregulation of the CXCL12/CXCR4 axis in two RT glioma cell lines (n=3). (B) Western blotting confirmed activation of the PI3K/mTOR pathway in two RT glioma cell lines (n=3). (C) Western blotting demonstrated enhanced autophagy in two RT glioma cell lines (n=3). (D) TEM revealed distinct alterations in autophagosomes and autolysosomes across two RT glioma cell lines (red arrows indicate autophagosomes and autolysosomes). Scale bars: 1 µm. ●, RT LN229 (RT U-87MG) group; ■, RT LN229 (RT U-87MG) + astrocytes culture medium group; ▲, RT LN229 (RT U-87MG) + RT-astrocytes culture medium group. *, P<0.05; **, P<0.01. RT, radiation‑treated; TEM, transmission electron microscopy.

Impact of recombinant CXCL12 on glioma cell viability and proliferation

To investigate the role of CXCL12, recombinant CXCL12 protein was used in the same experiments, establishing concentration gradients of 0, 2.5, 5, and 10 ng/mL. After treating RT LN229 (RT U-87MG) cells with different concentrations of CXCL12 for 24 hours. A 24-h CCK-8 assay revealed no significant difference in cell viability between the 0 and 2.5 ng/mL groups. However, when comparing the 0, 5, and 10 ng/mL groups, cell viability increased in a concentration-dependent manner. In LN229 cell experiments, no significant difference was identified between the 5 and 10 ng/mL treatment groups. In contrast, these two concentrations induced distinct responses in U-87MG cells. Consequently, the 2.5 ng/mL group was excluded from subsequent experiments, and the 10 ng/mL concentration did not further increase, as it was considered sufficiently high for the study’s purposes (Figure 4A). In subsequent experiments, as CXCL12 concentration increased, a corresponding elevation was found in Ki67 expression level (Figure 4B). Similarly, a reduction in apoptosis rate, decreased accumulation of ROS, and diminished DNA damage were identified (Figure 4C-4F).

Figure 4 The CXCL12 recombinant protein enhances cell proliferation, suppresses apoptosis, reduces ROS accumulation, and promotes radio-resistance and survival in RT-LN229 and RT-U87MG cells. (A) CCK-8 assay results at 24 hours for RT LN229 and RT U-87MG cells treated with varying concentrations of CXCL12 recombinant protein (n=3). (B) IF assay was used for detecting the Ki67 (n=3). (C) Flow cytometry was used to detect apoptosis (n=3). (D) Flow cytometry was used to detect ROS (n=3). (E) IF assay was employed for detecting the γ-H2AX (n=3). (F) IF assay was used for detecting the 8-OHdG (n=3). Scale bars: 20 µm. ns, not significant, *, P<0.05; **, P<0.01; ***, P<0.001. CCK-8, Cell Counting Kit-8; DAPI, 4’,6-diamidino-2-phenylindole; IF, immunofluorescence; ROS, reactive oxygen species; RT, radiation‑treated.

CXCL12-mediated suppression of autophagy in glioma cells

Irradiated glioma cells cultured in media with varying concentrations of CXCL12 exhibited a concentration-dependent elevation in the expression levels of CXCL12/CXCR4 signaling pathway-associated proteins (Figure 5A). Consistently, as the CXCL12 concentration rose, there was a marked elevation in the phosphorylation ratios of p-PI3K/PI3K and p-mTOR/mTOR, suggesting activation of the PI3K/mTOR signaling pathway (Figure 5B). Furthermore, Western blot analysis of autophagy-related proteins revealed that Beclin-1 expression level and LC3 II/LC3 I ratio were reduced, indicating that CXCL12 could suppress autophagy in irradiated glioma cells (Figure 5C).

Figure 5 The CXCL12 recombinant protein upregulates the CXCL12/CXCR4 chemokine axis in RT LN229 and RT U-87MG cells subjected to radiotherapy, subsequently activating autophagy through PI3K/mTOR pathway. (A) Western blotting confirmed upregulation of the CXCL12/CXCR4 axis in two RT glioma cell lines (n=3). (B) Western blotting confirmed activation of the PI3K/mTOR pathway in two RT glioma cell lines (n=3). (C) Western blotting demonstrated enhanced autophagy in two RT glioma cell lines. *, P<0.05; **, P<0.01. RT, radiation‑treated.

CXCL12 antibody interference reduced glioma cell viability and enhanced radio-sensitivity

To further validate the findings, an experiment was designed in which irradiated glioma cells were cultured in media derived from RT-astrocytes, with the addition of either a CXCL12 antibody or an IgG control. The findings revealed that the inclusion of the CXCL12 antibody led to a significant decrease in cell viability, whereas the IgG group remained unaffected (Figure 6A). Similarly, treatment with the CXCL12 antibody resulted in reduced Ki67 expression level, increased apoptosis, and elevated levels of accumulated ROS in the irradiated glioma cells, while the IgG group exhibited no such effects (Figure 6B-6D). Furthermore, the addition of the CXCL12 antibody resulted in the increased accumulation of markers of DNA damage in the cells, indicating a weakened radio-resistance in the irradiated glioma cells (Figure 6E,6F).

Figure 6 CXCL12 in RT-astrocyte-derived culture medium enhances cell proliferation, suppresses apoptosis, reduces ROS accumulation, and promotes radio-resistance and survival in RT-LN229 and RT-U87MG cells. (A) CCK-8 assay results at 24 hours for RT LN229 and RT U-87MG cells (n=3). (B) IF assay was used for detecting the Ki67 (n=3). (C) Flow cytometry was used to detect apoptosis (n=3). (D) Flow cytometry was used to detect ROS (n=3). (E) IF assay was employed for detecting the γ-H2AX (n=3). (F) IF assay was used for detecting the 8-OHdG (n=3). Scale bars: 20 µm. ns, not significant; *, P<0.05; **, P<0.01; ***, P<0.001. CCK-8, Cell Counting Kit-8; DAPI, 4’,6-diamidino-2-phenylindole; IF, immunofluorescence; ROS, reactive oxygen species; RT, radiation‑treated.

Effects of CXCL12 antibody on CXCL12/CXCR4 signaling and autophagy in glioma cells

When irradiated glioma cells were cultured in media derived from RT-astrocytes, the introduction of a CXCL12 antibody led to a notable reduction in the expression levels of CXCL12/CXCR4 signaling pathway-associated proteins in the antibody-treated irradiated glioma cells (Figure 7A). A similar downward trend was identified in the phosphorylation ratios of p-PI3K/PI3K and p-mTOR/mTOR, suggesting inhibition of the PI3K/mTOR signaling pathway (Figure 7B). Concurrently, Western blot analysis of autophagy-related proteins revealed an increase in Beclin-1 expression level and an elevated LC3 II/LC3 I ratio, indicating a diminished inhibitory effect on the autophagy pathway (Figure 7C). Moreover, TEM images demonstrated an increased number of autophagosomes and autolysosomes (Figure 7D).

Figure 7 CXCL12 in RT-astrocyte-derived culture medium activates the CXCL12/CXCR4 axis in RT LN229 and RT U-87MG cells, leading to autophagy modulation through the PI3K/mTOR pathway. (A) Western blotting confirmed activation of the CXCL12/CXCR4 axis in two RT glioma cell lines (n=3). (B) Western blotting confirmed activation of the PI3K/mTOR pathway in two RT glioma cell lines (n=3). (C) Western blotting demonstrated enhanced autophagy in two RT glioma cell lines (n=3). (D) TEM revealed distinct alterations in autophagosomes and autolysosomes across two RT glioma cell lines (red arrows indicate autophagosomes and autolysosomes). Scale bars: 1 µm for TEM. a, RT LN229 (U-87MG) + RT-astrocyte; b, RT LN229 (U-87MG) + RT-astrocyte + CXCL12 antibody; c, RT LN229 (U-87MG) + RT-astrocyte + IgG. ns, not significant; *, P<0.05; **, P<0.01. RT, radiation‑treated; TEM, transmission electron microscopy.

Tumor growth and radio-resistance in mouse xenograft models

In the animal experiments, tumor xenografts were established in nude mice by subcutaneously injecting RT-LN229 cells. Mice were divided into three groups: one group received RT-LN229 cells alone, another group was injected with RT-LN229 cells co-cultured with astrocytes, and the third group received RT-LN229 cells co-cultured with radiation-treated astrocytes (RT-astrocytes). Four mice were randomly assigned to each group. On day 15 post-tumor modeling, when the subcutaneous tumors had reached a detectable size and their stability was confirmed, a total radiation dose of 10 Gy was administered to the tumor region of each mouse. Mice were euthanized two days following the completion of radiotherapy (Figure 8A). It was revealed that tumor formation was challenging with RT-LN229 cells alone, as only one mouse developed a tumor. Consequently, this group was excluded from subsequent analyses. Notably, the tumor volumes in mice from the RT-LN229 + RT-astrocyte co-culture group were significantly elevated (Figure 8B), and murine tumor weights also significantly increased (Figure 8C). These findings demonstrate that the RT-LN229 + RT-astrocyte group exhibited enhanced tumor proliferation capacity and radio-resistance. Immunohistochemical analysis further confirmed these findings, revealing a higher Ki67 expression level, a lower Caspase-3 expression level, and reduced DNA damage markers in the tumors of the RT-LN229 + RT-astrocyte group (Figure 8D). Additionally, we measured only the maximum diameter (1.5 cm) and weight (0.9155 g) of the tumor. Using the spherical volume formula, the theoretical maximum volume was calculated as 1.77 cm³ based on the 1.5 cm diameter. However, due to the tumor’s irregular shape (e.g., ellipsoidal or lobulated), its actual volume should be smaller than this theoretical estimate.

Figure 8 Establishment of a subcutaneous tumor model to demonstrate the existence of radioresistance. (A) Schematic diagram of subcutaneous tumor experiment. (B) Gross appearance of the tumor (n=4). (C) Tumor weight (n=4). (D) Immunohistochemistry of Ki67, Caspase-3, γ-H2AX, and 8-OHdG in tumor tissues. Scale bars: 40 µm. **, P<0.01; ***, P<0.001. RT, radiation‑treated.

Immune infiltration and CXCL12 expression level in the TME

As a chemokine, does the high expression of CXCL12 affect immune infiltration in glioma? We delved into the disparities in immune infiltration attributed to CXCL12 through the following method: Utilizing the median expression level as a cutoff, we categorized the CGGA glioma transcriptomic dataset into two cohorts: one with high CXCL12 expression and the other with low expression. We then employed the CIBERSORT algorithm to assess the infiltration levels of 22 distinct immune cell subsets within the tumor tissues. The analysis revealed significant heterogeneity in the tumor immune microenvironment between the two groups, with significantly higher infiltration of CD8⁺ cytotoxic T lymphocytes, TAMs, and dendritic cells in the high-CXCL12 cohort (Figure 9A). As our previous experiments have demonstrated, radiotherapy can lead to an increase in CXCL12 within the microenvironment, which in turn contributes to the immune infiltration of gliomas following radiotherapy. To further confirm these findings, an animal experiment was designed. Given the need to observe immune responses, no human-derived glioma cells were utilized for tumor modeling in nude mice. A transplantable tumor model was therefore established using GL261 cells in C57BL/6 mice. Four distinct groups of mice were included: a normal control group (no treatment), a radiotherapy-alone group (single 10 Gy radiation dose to the head), a primary tumor model group (transplantable tumors generated using RT-GL261 cells), and a radiotherapy-relapsed glioma model group (mice first received a 10 Gy radiation dose, followed by transplantable tumor establishment). On the 15th day post-modeling, MRI was employed to monitor and assess tumor volumes (the upper portion of Figure 9B). Six mice were randomly assigned to each group.

Figure 9 CXCL12 expression stratifies distinct immune cell infiltration profiles and guides the design of subsequent in vivo experimental validation. (A) The immune cell profiles in subgroups stratified by high and low CXCL12 expression levels. (B) Schematic diagram of animal experiments. ns, not significant; *, P<0.05; **, P<0.01. MRI, magnetic resonance imaging.

Effectiveness of PD-1 antibody therapy in glioma models with radiotherapy relapse

Mice in the normal control group and those subjected solely to radiotherapy exhibited long-term survival, whereas the survival time of mice in the radiotherapy-relapsed glioma model group was significantly reduced compared with that of mice bearing primary tumors (Figure 10A). MRI examination revealed that intracranial tumors in mice from the radiotherapy-relapsed glioma group were significantly larger (Figure 10B). Immunohistochemical analysis was carried out on tumor-bearing brain tissues from both groups, including regions both surrounding and within the tumors. The markers analyzed included CD4, CD8, CD56, CD68, PD1, and Ki67 to assess the infiltration patterns of CD4⁺ T cells, CD8⁺ T cells, NK cells, and macrophages, along with the immunosuppressive status of immune cells and the proliferative activity of the tumors (Figure 10C). The results indicated that radiotherapy increased macrophage infiltration at the tumor periphery, but did not enhance the infiltration of CD4⁺ T cells, CD8⁺ T cells, or NK cells. The augmented presence of macrophages may be associated with the repair processes following radiotherapy-induced damage. In the periphery of primary glioma tumors, infiltration of NK cells and macrophages was noteworthy, a phenomenon that could be attributed to the tumor itself. However, in the recurrent tumor group, a significant increase in the infiltration of CD4⁺ T cells, CD8⁺ T cells, NK cells, and macrophages was evident at the tumor periphery, especially compared with the other groups. Furthermore, PD-1 expression level was significantly elevated in this group. In the tumor parenchyma, the recurrent tumor group exhibited significantly higher infiltration of CD4⁺ T cells, CD8⁺ T cells, and macrophages compared with the primary tumor group, whereas the difference in NK cell infiltration did not reach statistical significance. Concurrently, a notably elevated PD-1 expression level was identified in the recurrent tumor group, suggesting an escalated state of immunosuppression. Similarly, the radiotherapy-relapsed gliomas exhibited significantly increased Ki67 expression level, indicating a higher level of cellular proliferation (Figure 10D). According to the results illustrated in Figure 10A-10D, a therapeutic experiment involving PD-1 antibody treatment was further designed. In both the primary tumor model and the radiotherapy-relapsed glioma model, mice received PD-1 antibody treatments on days 6, 9, 12, and 15. Subsequently, on the evening of day 15, MRI scans were conducted to evaluate tumor sizes (the lower portion of Figure 9B). Six mice were allocated to each group for the experiment. The findings revealed that the primary tumor group exhibited a robust response to PD-1 antibody treatment, whereas the radiotherapy-relapsed tumor group demonstrated a markedly poor therapeutic outcome (Figure 10E,10F). The results of the LDH assay revealed that, in the co-culture system of effector cells (CD8⁺ T cells) and target cells (GL261 cells), the killing capacity of the mixed group consisting of CD8⁺ T cells and GL261 cells was stronger than that of the mixed group comprising CD8⁺ T cells, GL261 cells, and 50 ng/mL CXCL12 recombinant protein. This finding suggests that high concentration CXCL12 may exert an inhibitory effect on the non-specific killing function of CD8⁺ T cells (Figure 10G).

Figure 10 Intracranial tumor transplantation experiment. (A) Survival duration of normal mice, mice subjected to cranial irradiation, mice with solely transplanted tumors, and mice with transplanted tumors following cranial irradiation (n=6). (B) The MRI scan clearly illustrates the size of the tumor, and the tumor is outlined by a red dashed line. (C) IHC of CD4, CD8, CD56, PD1, and Ki67 in peritumoral tissue and tumor tissue. a, brain tissue from the normal mouse group; b, brain tissue from the mouse group subjected to radiotherapy; c, peritumoral brain tissue from the mouse group with tumor induction alone; d, peritumoral brain tissue from the mouse group with tumor induction followed by radiotherapy; e, tumor tissue from the mouse group with tumor induction alone; f, tumor tissue from the mouse group with tumor induction followed by radiotherapy. (D) Bar chart of immunohistochemistry (n=6). (E) Survival time of mice with solely transplanted tumors and mice with transplanted tumors following cranial irradiation after receiving anti-PD-1 therapy (n=6). (F) The MRI scan clearly illustrates the size of the tumor, and the tumor is outlined by a red dashed line. (G) Kill rate analysis of GL261 glioma cells co-cultured with CD8⁺ T cells. Scale bars: 40 µm for IHC images. ns, not significant; *, P<0.05; **, P<0.01; ***, P<0.001. IHC, immunohistochemistry; MRI, magnetic resonance imaging; PD-1, programmed cell death protein 1.

Discussion

This study elucidated a multifaceted mechanism by which radiotherapy dynamically reshapes the glioma microenvironment to drive adaptive resistance and modulate immunotherapeutic responsiveness. The results revealed that radiotherapy could induce astrocytes to secrete CXCL12, promoting the proliferation, survival, and radio-resistance of irradiated glioma cells through PI3K/mTOR-mediated suppression of autophagy, thereby creating a pro-tumorigenic niche. This suggests a distinct immunotherapeutic window in the post-radiation setting, despite the general trend of reduced treatment responsiveness in recurrent gliomas. These findings provide new insights into radiotherapy-driven tumor progression and highlight the potential of targeting microenvironmental adaptations to overcome treatment resistance in gliomas.

The TME in glioma represents a highly orchestrated ecosystem where stromal cells, including reactive astrocytes, TAMs, resident microglia, and neuronal elements, collectively drive malignant progression through dynamic paracrine networks. There is substantial evidence that bidirectional neuronal-glioma signaling can promote glioma growth through paracrine factors and direct electrochemical communication. Venkatesh et al. identified several secreted neuronal activity-dependent mitogens, including brain-derived neurotrophic factor (BDNF) and neuroligin-3 (NLGN3), both of which could promote glioma proliferation (22). Using in vivo xenograft models for HGGs, neuronal activity-dependent tumor cell depolarization increased glioma cell proliferation and promoted neurite-like protrusion formation and invasion (23,24). There is an AMPA receptor-mediated glutamatergic synapse between neurons and glioma cells. Glutamate released by presynaptic neurons can directly activate AMPA receptors (glur1/2) on the surface of glioma cells, triggering calcium ion influx and driving tumor cell proliferation and invasion (25). Tumor-associated reactive astrocytes can contribute to extracellular glutamate levels due to the reduced uptake (26). In a murine glioma model, Cx43-mediated gap junctions between non-neoplastic astrocytes and glioma cells promote tumor invasion (27,28). Reactive astrocytes can also promote an immunosuppressive microenvironment via the release of anti-inflammatory cytokines, such as transforming growth factor beta (TGF-β), IL-10, and granulocyte colony-stimulating factor (G-CSF) (19). Astrocyte secretion of chemokines may also directly modify tumor cells. For instance, CCL20 released from astrocytes can promote glioma adaptation to hypoxic stress by promoting hypoxia-inducible factor 1 alpha (HIF-1α) expression level (29). In some contexts, glioma-microglia cross-talk may be critical to promote tumor development (30). Among glioma-associated microglia, those exhibiting an M2 phenotype activate protein kinase D via secretion of IL-1β, which triggers the PI3K pathway, leading to phosphorylation of threonine 10 on glycerol-3-phosphate dehydrogenase. This subsequently enhances glycolytic flux and promotes tumor cell proliferation (31). Targeting glioma-associated microglia, by modulating their immunosuppressive functions, metabolic reprogramming, and interactions within the TME, has emerged as a highly promising novel target for immunotherapy in GBM (32). Unlike the homeostatic roles of stromal cells in the aforementioned physiological or pathological states, this study focuses on a specific therapeutic intervention—radiotherapy. Radiotherapy can specifically activate astrocytes into a type of key cell that supports tumor survival. Its core mechanism does not rely on the aforementioned known factors; instead, it functions by continuously secreting CXCL12 and acting on the specific axis of PI3K/mTOR-autophagy. This reveals a novel pathway for TME remodeling under therapeutic stress.

Tumor-educated TAMs release paracrine factors that promote glioma growth, survival, and invasion. Paracrine factors released by TAMs that can promote tumor growth include IL-1β, TGF-β1, stress-inducible protein 1 (STI-1), IL-10, and CCL5 (33-37). Our findings, in this context, underscore the unique position of astrocytes within the microenvironment following radiotherapy. In the setting of radiotherapy, the pro-survival role played by astrocytes through CXCL12 may coexist with or synergize with the effects of other stromal cells, such as TAMs, jointly constructing a more complex drug-resistant niche.

CXCL12 has been studied in both tumorous and non-tumorous diseases. Experiments by Julian Leberzammer have indicated that targeting platelet-derived CXCL12 could prevent arterial thrombosis (38). Numerous studies have found that CXCL12 was associated with tumor progression. CXCL12 promotes the invasion and metastasis of ovarian cancer cells by inducing epithelial-mesenchymal transition (EMT) via the PI3K/Akt signaling pathway (39). Previous research has found that circDLG1 interacted with miR-141-3p and acted as a miRNA sponge to increase CXCL12 expression level, which promoted gastric cancer progression (40). In colorectal cancer, increased kynurenine activity could inhibit KLF5-dependent CXCL12 expression level, thereby enhancing the immunosuppressive TME, improving the efficacy of immunotherapy and reducing metastasis (41). There is limited research on the relationship between radiotherapy-induced astrocyte changes and recurrent gliomas. In a study by Ji J et al., it was found that radiotherapy-induced astrocyte senescence could play a pivotal role in glioma recurrence. The underlying mechanism involves the release of the senescence-associated secretory phenotype (SASP), which subsequently remodels the tumor immune microenvironment (42). Notably, CXCL12 was also identified among the altered cytokine secretion profiles of the senescent astrocytes. This is consistent with our experiments, prompting us to conduct more in-depth research on the downstream signaling mechanisms of CXCL12.

While the present study utilized in vitro co-culture systems and head-irradiated xenograft models to investigate RT-induced glioma microenvironment remodeling, these approaches exhibited critical limitations: (I) microenvironment fidelity: preclinical models inadequately replicate the dynamic spatiotemporal changes observed clinically, including blood-brain barrier disruption kinetics, sequential immune cell infiltration patterns, and neurovascular unit remodeling; (II) treatment complexity: the xenograft system neglects confounding factors in human patients, such as surgical trauma-induced sterile inflammation, temozolomide-mediated chemo-modulation of tumor cell metabolism, and therapy-driven clonal evolution, all of which may differentially regulate CXCL12 secretion and autophagic flux in recurrent gliomas; (III) mechanistic resolution: upstream drivers of RT-induced astrocyte CXCL12 production remain uncharacterized, particularly the roles of NF-κB/HIF-1α signaling pathways and glutaminolysis-dependent metabolic reprogramming in mediating this pro-tumorigenic shift; (IV) while our in vitro data establish a direct causal role for astrocyte-derived CXCL12, the in vivo evidence remains correlative. Definitive proof of causality would require an in vivo intervention, such as CXCL12 blockade. However, interpreting such an experiment is complex because CXCL12 in the recurrent glioma microenvironment is produced by both activated astrocytes and the tumor cells themselves. Thus, systemic blockade would confirm the overall pathogenic role of CXCL12 but could not delineate the specific contribution of the stromal (astrocytic) source that is the focus of our mechanistic model. Future studies employing cell-type-specific genetic tools (e.g., conditional knockout of CXCL12 in astrocytes versus glioma cells) will be necessary to dissect these distinct contributions.

The experimental data present several aspects that merit thorough investigation. Initially, both CXCL12 ELISA and Western blot analyses focusing on proteins within the CXCL12/CXCR4 axis revealed a significant upregulation of CXCL12 expression in RT LN229 and RT U-87MG glioma cells. Importantly, when CXCL12 secreted by irradiated astrocytes interacted with irradiated glioma cells, it appeared to stimulate the release of additional CXCL12 from the glioma cells. Further analysis demonstrated that irradiated glioma cells themselves were capable of secreting substantial amounts of CXCL12. However, in the context of recurrent gliomas, the tumor cell population at this stage is markedly reduced, limiting their ability to establish a clinically relevant concentration of CXCL12. In contrast, irradiated astrocytes persist in high numbers during this phase. These astrocytes not only provide residual glioma cells with the required CXCL12 concentration but also seem to enhance the autocrine secretion of CXCL12 by the glioma cells themselves, thereby amplifying the CXCL12-mediated signaling within the TME.

In clinical practice, a subset of patients afflicted with recurrent gliomas are deprived of surgical intervention opportunities and consequently resort to a treatment regimen comprising radiotherapy combined with immunotherapy. Nevertheless, drawing upon the insights gleaned from our experimental findings, it is imperative to prioritize the practical application of immunotherapy in the treatment of primary tumors. In contrast, the strategy of introducing immunotherapy as a salvage therapy after tumor recurrence should not be the focal point of the treatment approach. Our experimental findings offer a mechanistic explanation for this viewpoint: Radiotherapy has created, at the recurrence site, a microenvironment highly conducive to tumor survival and potentially immunosuppressive, dominated by activated astrocytes and the CXCL12 feedback loop. This may partly explain the poor response to immunotherapy following recurrence. Therefore, intervening in this pathway (such as combining CXCL12/CXCR4 inhibitors) concurrently with or immediately after the initial treatment (radiotherapy) may hold a strategic advantage over waiting to employ immunotherapy until after recurrence.

In conclusion, the present study revealed a dual mechanism of RT-induced adaptive resistance: astrocyte-glioma interaction could promote survival, and recurrent tumors exhibited poorer immunotherapy outcomes. These findings expand the scope of radiobiology and provide a theoretical basis for targeting microenvironment-driven resistance, highlighting transformative potential for optimizing glioma treatment paradigms.

Conclusions

These findings suggest that immunotherapy should be prioritized for the treatment of primary tumors rather than being added to regimens for recurrent tumors, where it has already demonstrated poor therapeutic efficacy.

Acknowledgments

We thank Medjaden Inc. for scientific editing of this manuscript.

Footnote

Reporting Checklist: The authors have completed the ARRIVE and MDAR reporting checklists. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2652/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2652/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2652/prf

Funding: The study was supported by the National Natural Science Foundation of China (Nos. 82460517 and 82360566); the Natural Science Foundation Of Gansu Province (Nos. 25YFFA054 and 2025RCXM015); the Lanzhou Science and Technology Bureau Project (No. 2021-RC-97/2023-1-48); Cuiying Graduate Guidance Teacher Cultivation Program Project of Lanzhou University Second Hospital (No. CYDSPY202002/YJS-BD-13); and the Natural Science Foundation of Ningxia Hui Autonomous Region (No. 2024AAC03472).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2652/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All animal experiments were performed under a project license (No. D2023-034) granted by the the Institutional Animal Care and Use Ethics Committee of Lanzhou University Second Hospital, in compliance with the ethical guidelines set by Lanzhou University Second Hospital for the care and use of animals.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Liu Z, Wu Y, Xu H, et al. Multimodal fusion of radio-pathology and proteogenomics identify integrated glioma subtypes with prognostic and therapeutic opportunities. Nat Commun 2025;16:3510. [Crossref] [PubMed]
2. Stępka J, Dotka M, Kosiński M, et al. The Role of Systemic Therapies in the Treatment of Grades 1-4 Gliomas. Cureus 2024;16:e70532. [Crossref] [PubMed]
3. Nabors LB, Portnow J, Ahluwalia M, et al. Central Nervous System Cancers, Version 3.2020, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw 2020;18:1537-70. [Crossref] [PubMed]
4. Harat M, Małkowski B, Wiatrowska I, et al. Relationship between Glioblastoma Dose Volume Parameters Measured by Dual Time Point Fluoroethylthyrosine-PET and Clinical Outcomes. Front Neurol 2017;8:756. [Crossref] [PubMed]
5. Weller M, Cloughesy T, Perry JR, et al. Standards of care for treatment of recurrent glioblastoma--are we there yet? Neuro Oncol 2013;15:4-27. [Crossref] [PubMed]
6. Drappatz J, Mantica M. Treatment strategies and innovation for recurrent high-grade glioma. J Neurooncol 2025;176:118. [Crossref] [PubMed]
7. Tsuji Y, Nonoguchi N, Okuzaki D, et al. Chronic pathophysiological changes in the normal brain parenchyma caused by radiotherapy accelerate glioma progression. Sci Rep 2021;11:22110. [Crossref] [PubMed]
8. Campos B, Olsen LR, Urup T, et al. A comprehensive profile of recurrent glioblastoma. Oncogene 2016;35:5819-25. [Crossref] [PubMed]
9. Sahai E, Astsaturov I, Cukierman E, et al. A framework for advancing our understanding of cancer-associated fibroblasts. Nat Rev Cancer 2020;20:174-86. [Crossref] [PubMed]
10. Zhang Q, Liu Y, Zhang Z, et al. GLUL Confers Perivascular Cancer-Associated Fibroblasts With Pro-Angiogenic Capacity to Promote Glioma Progression. Adv Sci (Weinh) 2026;13:e13184. [Crossref] [PubMed]
11. Frisbie L, Buckanovich RJ, Coffman L. Carcinoma-Associated Mesenchymal Stem/Stromal Cells: Architects of the Pro-tumorigenic Tumor Microenvironment. Stem Cells 2022;40:705-15. [Crossref] [PubMed]
12. Sun L, Cao X, Zhou B, et al. Tumor-associated mesenchymal stem/stromal cells in tumor microenvironment and carcinogenesis. Exp Hematol Oncol 2025;14:97. [Crossref] [PubMed]
13. Öhlund D, Elyada E, Tuveson D. Fibroblast heterogeneity in the cancer wound. J Exp Med 2014;211:1503-23. [Crossref] [PubMed]
14. Zhang L, Wang K, Zhang J, et al. PDGFR-β/Cav1-induced autophagy via mTOR/FIP200/ATG13 activation in cancer-associated fibroblasts promotes the malignant progression of breast cancer. J Transl Med 2025;23:784. [Crossref] [PubMed]
15. Liu Q, Yu M, Lin Z, et al. COL1A1-positive endothelial cells promote gastric cancer progression via the ANGPTL4-SDC4 axis driven by endothelial-to-mesenchymal transition. Cancer Lett 2025;623:217731. [Crossref] [PubMed]
16. Endo F, Kasai A, Soto JS, et al. Molecular basis of astrocyte diversity and morphology across the CNS in health and disease. Science 2022;378:eadc9020. [Crossref] [PubMed]
17. Ferris HA, Perry RJ, Moreira GV, et al. Loss of astrocyte cholesterol synthesis disrupts neuronal function and alters whole-body metabolism. Proc Natl Acad Sci U S A 2017;114:1189-94. [Crossref] [PubMed]
18. Perelroizen R, Philosof B, Budick-Harmelin N, et al. Astrocyte immunometabolic regulation of the tumour microenvironment drives glioblastoma pathogenicity. Brain 2022;145:3288-307. [Crossref] [PubMed]
19. Henrik Heiland D, Ravi VM, Behringer SP, et al. Tumor-associated reactive astrocytes aid the evolution of immunosuppressive environment in glioblastoma. Nat Commun 2019;10:2541. [Crossref] [PubMed]
20. Priego N, Zhu L, Monteiro C, et al. STAT3 labels a subpopulation of reactive astrocytes required for brain metastasis. Nat Med 2018;24:1024-35. [Crossref] [PubMed]
21. Sofroniew MV. Astrocyte Reactivity: Subtypes, States, and Functions in CNS Innate Immunity. Trends Immunol 2020;41:758-70. [Crossref] [PubMed]
22. Venkatesh HS, Johung TB, Caretti V, et al. Neuronal Activity Promotes Glioma Growth through Neuroligin-3 Secretion. Cell 2015;161:803-16. [Crossref] [PubMed]
23. Venkatesh HS, Morishita W, Geraghty AC, et al. Electrical and synaptic integration of glioma into neural circuits. Nature 2019;573:539-45. [Crossref] [PubMed]
24. Minns HE, Useinovic N, Smith JL, et al. Electrophysiology of gliomas: current science, implications, and opportunities. Front Oncol 2025;15:1611840. [Crossref] [PubMed]
25. Venkataramani V, Tanev DI, Strahle C, et al. Glutamatergic synaptic input to glioma cells drives brain tumour progression. Nature 2019;573:532-8. [Crossref] [PubMed]
26. Campbell SC, Muñoz-Ballester C, Chaunsali L, et al. Potassium and glutamate transport is impaired in scar-forming tumor-associated astrocytes. Neurochem Int 2020;133:104628. [Crossref] [PubMed]
27. Song Y, Huang Q, Pu Q, et al. Gastrodin Liposomes Block Crosstalk between Astrocytes and Glioma Cells via Downregulating Cx43 to Improve Antiglioblastoma Efficacy of Temozolomide. Bioconjug Chem 2024;35:1380-90. [Crossref] [PubMed]
28. Sin WC, Aftab Q, Bechberger JF, et al. Astrocytes promote glioma invasion via the gap junction protein connexin43. Oncogene 2016;35:1504-16. [Crossref] [PubMed]
29. Jin P, Shin SH, Chun YS, et al. Astrocyte-derived CCL20 reinforces HIF-1-mediated hypoxic responses in glioblastoma by stimulating the CCR6-NF-κB signaling pathway. Oncogene 2018;37:3070-87. [Crossref] [PubMed]
30. Jayaram MA, Phillips JJ. Role of the Microenvironment in Glioma Pathogenesis. Annu Rev Pathol 2024;19:181-201. [Crossref] [PubMed]
31. Lu J, Xu Z, Duan H, et al. Tumor-associated macrophage interleukin-β promotes glycerol-3-phosphate dehydrogenase activation, glycolysis and tumorigenesis in glioma cells. Cancer Sci 2020;111:1979-90. [Crossref] [PubMed]
32. Wang B, Li C, Gu J, et al. Targeting glioma-associated microglia and macrophages: a new frontier in glioblastoma immunotherapy. Front Immunol 2025;16:1726440. [Crossref] [PubMed]
33. Solga AC, Pong WW, Kim KY, et al. RNA Sequencing of Tumor-Associated Microglia Reveals Ccl5 as a Stromal Chemokine Critical for Neurofibromatosis-1 Glioma Growth. Neoplasia 2015;17:776-88. [Crossref] [PubMed]
34. Chen J, Zhu L, Lawson T, et al. Role of elevation of glycolysis in tumor-associated macrophages in glioblastoma immune evasion and therapeutic implications. PNAS Nexus 2026;5:pgaf396. [Crossref] [PubMed]
35. Friedrich M, Sankowski R, Bunse L, et al. Tryptophan metabolism drives dynamic immunosuppressive myeloid states in IDH-mutant gliomas. Nat Cancer 2021;2:723-40. [Crossref] [PubMed]
36. Liu H, Sun Y, Zhang Q, et al. Pro-inflammatory and proliferative microglia drive progression of glioblastoma. Cell Rep 2021;36:109718. [Crossref] [PubMed]
37. Carvalho da Fonseca AC, Wang H, Fan H, et al. Increased expression of stress inducible protein 1 in glioma-associated microglia/macrophages. J Neuroimmunol 2014;274:71-7. [Crossref] [PubMed]
38. Leberzammer J, Agten SM, Blanchet X, et al. Targeting platelet-derived CXCL12 impedes arterial thrombosis. Blood 2022;139:2691-705. [Crossref] [PubMed]
39. Yu X, Shi L, Qi Z, et al. CXCL12 promotes EMT-mediated malignant transformation of endometriosis-associated ovarian cancer via PI3K/Akt signaling: An integrated transcriptomic and clinical study. Biol Reprod 2026;ioag016.
40. Chen DL, Sheng H, Zhang DS, et al. The circular RNA circDLG1 promotes gastric cancer progression and anti-PD-1 resistance through the regulation of CXCL12 by sponging miR-141-3p. Mol Cancer 2021;20:166. [Crossref] [PubMed]
41. Wei R, Zhou Y, Li C, et al. Ketogenesis Attenuates KLF5-Dependent Production of CXCL12 to Overcome the Immunosuppressive Tumor Microenvironment in Colorectal Cancer. Cancer Res 2022;82:1575-88. [Crossref] [PubMed]
42. Ji J, Ding K, Cheng B, et al. Radiotherapy-Induced Astrocyte Senescence Promotes an Immunosuppressive Microenvironment in Glioblastoma to Facilitate Tumor Regrowth. Adv Sci (Weinh) 2024;11:e2304609. [Crossref] [PubMed]