Human U87MG, U251MG, U373MG, and T98G GBM cell lines, and murine microglial BV-2 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) 4.5 g/L glucose (Sigma Aldrich, St. Louis, MO) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Sigma Aldrich, St. Louis, MO), 1% L-glutamine (Corning, Manassas, VA), 1% penicillin and streptomycin solution (Corning, Manassas, VA), and 1% amphotericin B (Sigma Aldrich, St. Louis, MO). Cells were cultured in a 5% CO₂ humidified atmosphere at 37 °C. Cells were maintained in T75 flasks (75 cm²) at a seeding density of 2 × 10⁶ cells per flask and subcultured three times per week upon reaching approximately 80% confluence. U87MG ATCC [Research Resource Identifier (RRID): CVCL_0022], U251MG (RRID: CVCL_0021), U373MG Uppsala (RRID: CVCL_2818), T98G (RRID: CVCL_0556), and BV-2 (RRID: CVCL_0182) cell lines were obtained from the Cell Bank, IRCCS AOU San Martino-IST (Genova, Italy). All cell lines were tested for mycoplasma contamination by polymerase chain reaction (PCR) every 3 weeks and were confirmed to be mycoplasma-free throughout the study. All cell lines were used between passages 1 and 5, except T98G, which has a stable profile and was used between passages 1 and 25.

Bone marrow from 6-8 week-old female C57BL/6J mice was isolated from femurs and tibias as reported in^[30] and cultured in Roswell Park Memorial Institute medium (RPMI) medium (ThermoFisher Scientific, Waltham, MA) supplemented with 5% heat-inactivated fetal calf serum, 20 ng/mL murine colony-stimulating factor (CSF) (UGent-IRC-VIB Protein Core Facility, Ghent, Belgium), 1% L-glutamine, and 50 μM 2-mercaptoethanol (ThermoFisher Scientific, Waltham, MA). In total, three animals (n = 3) were used as three independent biological replicates. The bone marrow was flushed with a 25 G needle (0.5 mm × 25 mm), resuspended, and coarse debris filtered through a CellStrainer 70 μm (ThermoFisher Scientific, Waltham, MA) and cleared from red blood cells using red blood cell (RBC) Lysis buffer 1X (ThermoFisher Scientific, Waltham, MA). Cells were cultured for up to 10 days. The culture medium was replaced with fresh medium on days 3, 6, and 9. All animal experiments were approved by the Animal Ethical Committee of the Faculty of Medicine and Health Sciences of Ghent University (approval/accreditation number: LA1400080; project reference: ECD 20/06aanp).

Cell viability was evaluated with the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay following the modified protocol of Sladowski et al.^[31]. U87MG, U251MG, U373MG, and T98G cells (1 × 10⁴ cells/100 μL/well) were seeded into 96-well plates and exposed to TMZ (100, 200, 400, and 800 μM) (ThermoFisher Scientific, Waltham, MA) for 24, 48, and 72 h. Cells were then incubated with 1 mg/mL of MTT (PanReac AppliChem, Darmstadt, Germany) in DMEM for 3 h at 37 °C and 5% CO₂, washed three times with phosphate-buffered saline (PBS; Aurogene, Rome, Italy) and lysed with dimethyl sulfoxide (DMSO; ThermoFisher Scientific, Waltham, MA) to dissolve the formazan crystals. Absorbance was read at 570 nm on a Multiskan^TM FC Microplate Photometer (ThermoFisher Scientific, Waltham, MA).

U87MG and U251MG cells (1 × 10⁴ cells/100 μL/well) were seeded into 96-well plates and exposed to TMZ (100, 200, 400, and 800 μM) for 48 h. Cells were then fixed with 4% paraformaldehyde (PFA; Sigma-Aldrich, St. Louis, MO) in PBS (30 min, RT), stained with 0.1% crystal violet (Sigma-Aldrich, St. Louis, MO) [1 h, room temperature (RT)], rinsed with water, and air-dried. Bound dye was solubilized in glacial acetic acid (Sigma-Aldrich, St. Louis, MO), and absorbance was measured at 550 nm with a Multiskan^TM FC Microplate Photometer.

EVs were isolated from conditioned media of cells treated with 200 μM TMZ for 48 h in EV-depleted 10% FBS-supplemented DMEM. After sequential centrifugation to remove cells and debris, the medium was concentrated using 10 kDa molecular weight cut-off (MWCO) Amicon Ultra-15 filters at 4,000 × g (Merck Millipore, Burlington, MA), and EVs were purified by SEC using a 10 mL Sepharose CL-2B column (Cytiva, Uppsala, Sweden), as reported in^[32]. EV-rich, protein-poor fractions (from 8 to 10) were collected in PBS and further concentrated to 100 μL with Amicon Ultra-4 filters (10 kDa MWCO; Merck Millipore, Burlington, MA) at 4,000 × g for 20 min.

EV size and concentration were assessed by NTA using a NanoSight® NS300 (Malvern Panalytical Ltd., Malvern, UK), equipped with a 488 nm blue laser and a high-sensitivity sCMOS camera. Recordings were analysed with NTA 3.4 Build 3.4.4 software (Malvern Panalytical Ltd., Malvern, UK). Samples were diluted 1:50 in freshly filtered PBS, vortexed for 1 min, and injected using a syringe pump at 30 arbitrary units (AU). For each of the three independent biological replicates, five 1-min videos were recorded and analysed to determine particle concentration (particles/mL), size distribution, mean, mode, standard deviation, diameter distributions (D10, D50, D90), and valid tracks.

For TEM analysis, untreated and TMZ-treated GBM cells were fixed for 1 h with 2.5% glutaraldehyde (Electron Microscopy Sciences, Hatfield, PA) in 0.1 M cacodylate buffer pH 7.4 (Sigma-Aldrich, St. Louis, MO) and post-fixed for 2 h with 1% osmium tetroxide (OsO₄; Electron Microscopy Sciences, Hatfield, PA) in the same buffer, dehydrated in ethanol, embedded in Spurr’s resin (Electron Microscopy Sciences, Hatfield, PA), sectioned into 60 nm slices (ultramicrotome RMC Boeckeler, Tucson, AZ), and placed on 200-mesh copper grids (Ted Pella, Inc., Redding, CA). For isolated EVs, samples were fixed with 0.1% PFA in PBS for 30 min at RT, loaded onto 200-mesh Formvar/carbon-coated grids (Ted Pella, Inc., Redding, CA), stained with 2% uranyl acetate (Electron Microscopy Sciences, Hatfield, PA, USA), and all samples were examined using a Zeiss Auriga Scanning Electron Microscope (Carl Zeiss Microscopy GmbH, Jena, Germany) equipped with the STEM module, operating at 20 keV.

Cryo-EM samples of EVs were prepared as described in^[33]. Briefly, a 4 μL aliquot of each EV sample was applied to glow-discharged grids, followed by blotting and plunge-freezing in liquid ethane. Glacios cryo-transmission electron microscope (ThermoFisher Scientific, Waltham, MA) was used to collect micrographs.

Fixed EV samples were loaded on 200-mesh Formvar/carbon-coated grids by inversion and then floated sequentially on droplets of PBS (three times for 3 min), 0.5% bovine serum albumin (BSA; Sigma-Aldrich, St. Louis, MO) in PBS (10 min), primary antibody anti-CD63 (60 min; 1:500 in PBS with 5% BSA, #sc-365604; Santa Cruz Biotechnology, Dallas, TX), 1% BSA (six times for 3 min), 20 nm protein A-gold conjugate (40 min; Abcam, Cambridge, UK; ab270601), 1% BSA (three times for 3 min) and, finally, washed in PBS (three times for 3 min). Then, grids were contrasted in 2% uranyl acetate (7 min, RT) and observed under a Zeiss Auriga Scanning Electron Microscope (Carl Zeiss Microscopy GmbH, Jena, Germany) equipped with the STEM module, operating at 20 keV.

AFM analysis of GBM cells and isolated EVs was performed as previously described by our group^[34], with minor adaptations. Briefly, GBM cells (5 × 10⁴ cells/well) were seeded on glass coverslips in 6-well plates, treated as required, and fixed on ice for 30 min with 1% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4), then air-dried at RT for 1 h. Isolated EVs were fixed with 0.1% PFA in PBS (30 min, RT) and diluted in Milli-Q water. Images were acquired in contact mode on a Multimode AFM (Bruker, Santa Barbara, CA) using DNP-model cantilever tips (Bruker; nominal spring constant 0.06 N/m, tip radius 10 nm), keeping the tip-sample force below 1 nN. Cells were scanned at 60 μm with 20 μm zoom-ins on areas of interest, while EVs were imaged at 20 μm and, more extensively, at 10 μm until enough vesicles were captured (scan speed 1-4 s/line, ≥ 512 points/line). Images were processed in Gwyddion (mean-plane subtraction, tilt correction, line-levelling), and vesicles were segmented and counted using the “Mark Grains by Segmentation” tool. Surface roughness (Rq) was calculated in ImageJ v1.8.0^[35] on five randomly selected 6 μm² × 6 μm² regions per sample.

Cells were lysed in radioimmunoprecipitation assay (RIPA) buffer (NaCl 150 mM, Tris-HCl 50 mM pH 8, MgCl₂ 2 mM, sodium dodecyl sulfate (SDS) 0.1%, deoxycholic acid 0.5%, NP-40 1%; Sigma-Aldrich, St. Louis, MO) containing phenylmethylsulfonyl fluoride (PMSF) and protease inhibitor cocktail (ThermoFisher Scientific, Waltham, MA). After sonication (5 min), insoluble material was pelleted by centrifugation for 30 min at 13,000 × g at 4 °C. Supernatants were transferred to a new tube, and the protein concentration was measured by Bradford assay using Bradford reagent (Bio-Rad Laboratories, Hercules, CA). 15-30 μg of protein was separated by SDS-polyacrylamide gel electrophoresis (10%-12% polyacrylamide; SureCastTM Acrylamide Solution; ThermoFisher Scientific, Waltham, MA) and transferred to nitrocellulose membranes (Bio-Rad Laboratories, Hercules, CA). For Western blot analysis of EVs, 10 μg of proteins from the EV-enriched fractions were used. The membranes were blocked with 5% non-fat dry milk (PanReac AppliChem, Darmstadt, Germany) and/or 3% BSA (blocking buffer) in Tris-buffered saline containing 0.1% Tween 20 (TBST; Sigma-Aldrich, St. Louis, MO) for 1 h at RT. Then, the membranes were incubated overnight at 4 °C with the following primary antibodies: tumor susceptibility gene 101 (TSG101) (1:1,000 dilution, #sc-7964; Santa Cruz Biotechnology, Dallas, TX), CD63 (1:1,000 dilution, #sc-365604; Santa Cruz Biotechnology, Dallas, TX), and ALG-2-interacting protein X (ALIX) (1:1,000 dilution, #sc-53540; Santa Cruz Biotechnology, Dallas, TX) diluted in blocking buffer. After washing with TBST, the membranes were incubated for 1 h at RT with a goat anti-mouse IgG Heavy and Light Chain secondary antibody (1:5,000, #A90-116P; Bethyl Laboratories, Montgomery, AL) diluted in blocking buffer. The immunoreactive bands were detected with a ChemiDoc imaging system (Bio-Rad Laboratories, Hercules, CA) using commercial enhanced chemiluminescence (ECL) reagent (WesternBrightTM Quantum, Advansta Inc., San Jose, CA). To check cell death, the membranes were incubated overnight at 4 °C with the following primary antibodies: caspase-3 (1:1,000 dilution, A19654, ABclonal, Wuhan, China), caspase-8 (1:1,000 dilution, A19549, ABclonal, Wuhan, China), caspase-9 (1:1,000 dilution, A18676, ABclonal, Wuhan, China), LC3B (1:1,000 dilution, A19665, ABclonal, Wuhan, China), glutathione peroxidase 4 (GPX4) (1:1,000 dilution, A11243, ABclonal, Wuhan, China), and β-actin (1:1,000 dilution, #bs-0061R, Bioss, Beijing, China) diluted in blocking buffer. To check BV-2 activation, the membranes were incubated with the following primary antibodies: CD80 (1:1,000 dilution, A16039, ABclonal, Wuhan, China) and CD206 (1:1,000 dilution, A11192, ABclonal, Wuhan, China). After washing with TBST, the membranes were incubated for 1 h at RT with a Goat anti-Rabbit IgG Heavy and Light Chain secondary antibody (1:5,000 dilution, #A120-101P, Bethyl Laboratories, Montgomery, AL) diluted in blocking buffer.

To profile the EV protein cargo, SEC-purified EVs from three independent biological replicates of T98G, U87MG, U251MG and U373MG cells were pooled, sonicated on ice (Sonicator U200S, 70% amplitude; IKA Labortechnik, Staufen, Germany) and processed by Filter-Aided Sample Preparation (FASP). Tryptic digests (15 μg protein/sample, Bradford-verified) were analysed in triplicate on a ultimate^TM 3000 ultra-high performance liquid chromatography (UHPLC) coupled to an Orbitrap Fusion^TM Tribrid^TM mass spectrometer (both ThermoFisher Scientific, Waltham, MA), following the chromatographic set-up previously described^[36], peptides loaded on a Trap Cartridge C18 (0.3 mm ID, 5 mm L, 5 μm PS) and resolved on an EASY-spray Acclaim^TM PepMap^TM C18 column (75 μm ID, 15 cm L, 2 μm PS; both ThermoFisher Scientific) at 300 nL/min over a 65-min run. Raw data were processed in MaxQuant 1.6.6.0 and Perseus 1.6.10.50 (Max Planck Institute for Biochemistry, Martinsried, Germany) against the UniProt Homo sapiens database (release 2020_06, 20,588 entries), as reported previously^[37]: cysteine carbamidomethylation was fixed, methionine oxidation and asparagine/glutamine deamidation were variable, and false discovery rate (FDR) was capped at 1% at both protein and peptide level. Protein abundance was estimated via intensity-based absolute quantification (iBAQ) values, requiring ≥ 2 valid values in at least one treatment group per comparison. Differential protein ratios (TMZ-treated vs. untreated) were analysed for gene ontology (GO) and functional enrichment using Ingenuity Pathway Analysis (IPA, Qiagen, Hilden, Germany; accessed February 2025), which predicts activation (z ≥ 2.0) or inhibition (z ≤ -2.0) of transcriptional regulators based on curated literature evidence^[38]. Proteomics data were deposited in the ProteomeXchange Consortium via PRoteomics IDEntifications database (PRIDE) (dataset PXD070357)^[39].

EVs were labelled with PKH26 Red Fluorescent Cell Linker Kits for General Cell Membrane Labeling (Sigma-Aldrich, St. Louis, MO) according to the manufacturer’s instructions. 1 mL EV sample (volume adjusted with diluent C) and 6 μL of PKH26 dye were mixed, gently pipetted for 30 s, and then left at RT for 5 min. The excess dye was removed by a second SEC purification. PKH26-only controls were processed, with PBS substituting the EV sample. 5 × 10⁵ of GBM cells, BMDCs, or BV-2 cells were then exposed to PKH26-labelled EVs (4 µg/mL), fixed at different time points (30 min, 1, 3, and 24 h) with 4% PFA, washed twice with PBS and blocked with 5% BSA in PBS, and then samples were stained for β-actin (1:200; AC004, ABclonal, Wuhan, China), washed twice with PBS and incubated with secondary Goat Anti-Mouse IgG H&L fluorescein isothiocyanate (FITC)-conjugated antibody (1:400; ab6785, Abcam, Cambridge, UK). Finally, nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI) Fluoromount-G (SouthernBiotech, Birmingham, AL). EV uptake was examined by confocal laser scanning microscopy (Leica Geosystems, Wetzlar, Germany) or Axio Imager M2 equipped with Apotome 2 (Zeiss GmbH, Jena, Germany). Z-stack imaging was used to distinguish internalized EVs from surface-bound EVs. Images were analysed using ImageJ Software version 1.54p^[34].

A total of 5 × 10⁵ cells per well of U87MG, U251MG cells, BMDCs, or BV-2 cells were seeded in 6-well plates and treated with 4 µg/mL of EVs in EV-depleted medium for 24 h. After the treatments, cells were processed for all subsequent functional analyses.

T98G, U87MG, U251MG, and U373MG cells were seeded at a density of 5 × 10⁵ cells/well into 6-well plates, and a scratch after 24 h was made in the centre of each well using a 200 μL pipette tip to assess migration. After washing out debris, cells were treated with 200 μM TMZ in 1% FBS-supplemented medium or with EVs from U87MG and U251MG cells in EV-depleted 1% FBS-supplemented medium. Wound closure was monitored over time (0, 1, 3, 6, 12, and 24 h, and additionally at 48 and 72 h for U251MG) by bright-field microscopy (Zeiss GmbH, Jena, Germany). Cell migration was quantified as gap area in at least three random fields using the MRI Wound Healing Tool plugin in ImageJ 1.53 software^[34].

To examine the effect of EVs derived from GBM cells on BMDC maturation, isolated EVs from both TMZ-treated and untreated U87MG and U251MG cells were added to the dendritic cells (DCs) starting from day 7 of culture for 24 h. Lipopolysaccharide (LPS; 1 μg/mL, Escherichia coli O111:B4; Sigma-Aldrich, St. Louis, MO) was added for an additional 24 h. LPS without EVs was used as a positive control, while untreated cells were used as a negative control. Both adherent and suspended cells cultured by the above methods were harvested from 6-well plates and centrifuged at 500 × g for 5 min at 4 °C. After centrifugation, the cells were washed and resuspended in fluorescence-activated cell sorting (FACS) buffer. Next, DCs were incubated for 20-30 min with the following antibodies: eFluor 450-conjugated anti-CD86 (clone GL1; eBioscience, San Diego, CA), Alexa Fluor 488-conjugated anti-major histocompatibility complex-II molecule (MHCII/IA-IE clone M5/114.15.2; BD Biosciences, San Jose, CA), eFluor 450-conjugated anti-CD80 (clone 16-10A1; eBioscience, San Diego, CA), allophycocyanin (APC)-conjugated anti-CD11c (clone N418, eBioscience, San Diego, CA), and mouse Fc-blocker (clone 93, eBioscience, San Diego, CA). After incubation, cells were washed with PBS by centrifugation at 300 × g for 5 min at 4 °C. The supernatant was removed, and the cells were resuspended in 200 µL of PBS. Flow cytometric analyses were performed with a BD LSR II flow cytometer (BD Biosciences, San Jose, CA) at the UGent Core Flow Cytometry facility of Ghent University. At least 10,000 events were recorded for each sample. In all the analyses, debris and clumps were gated out. The data were analysed with FlowJo Flow Cytometry Analysis Software (TreeStar Inc., Ashland, OR).

Total RNA was extracted by TRI-TIZ G (PanReac AppliChem, Darmstadt, Germany), and 1 μg RNA was used to generate complementary DNA using the SensiFAST complementary DNA (cDNA) synthesis kit (Meridian Bioscience, London, UK) according to the manufacturers’ protocols. The resulting cDNA was used for qPCR with SensiFAST Probe Lo-ROX kit (Meridian Bioscience, London, UK) with the following primer pairs: Tbp (Forward primer TGGAATTGTACCGCAGCTTCAAAAT; Reverse primer GCAGTTGTCCGTGGCTCTCT), IL-1β (Forward primer TGCCACCTTTTGACAGTGATGA; Reverse primer TGCCTGCCTGAAGCTCTTGT), IL-10 (Forward primer TAGAAGTGATGCCCCAGGCAG; Reverse primer ACACCTTGGTCTTGGAGCTTATT), and analysed with QuantStudio^TM 3 Real-Time PCR System (Applied Biosystems, ThermoFisher Scientific, Waltham, MA). Relative gene expression levels were calculated with the ΔΔCT method.

All data are expressed as mean ± standard deviation (SD) of at least three indipendent biological replicates. Multiple comparisons were performed using either one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test or by two-way ANOVA. Comparisons between two groups were performed using Student’s t-test. Statistical analyses were performed using GraphPad Prism 7 (GraphPad Software, San Diego, CA), and P < 0.05 was considered statistically significant.

Firstly, TEM analysis was used to identify MVBs and budding EVs in all four human GBM cell lines. MVBs with ILVs and plasma membrane budding were observed in all GBM cell types, with no significant differences in the size or number of MVBs [Figure 1A]. Next, EV-enriched fractions from the four GBM cell lines were isolated using UF/SEC. EV morphology was assessed using TEM, cryo-EM, and AFM [Figure 1B-D]. TEM micrographs revealed EVs with a characteristic round, cup-shaped morphology, with sizes ranging from 20 to 160 nm for T98G, U87MG, and U373MG-derived EVs, and from 10 to 180 nm for U251MG-derived EVs [Figure 1B and Supplementary Figure 1A]. Cryo-EM analysis revealed EVs in a near-native morphological state, predominantly round and enclosed by clearly visible lipid bilayers [Figure 1C]. Importantly, no round, moderately electron-dense lipoproteins lacking a lipid bilayer were detected. This confirms the success of the isolation protocol^[33]. In addition, a well-established AFM protocol from our group was used^[35], which corroborated these morphological features [Figure 1D] and provided more statistically reliable information. Specifically, AFM-based analysis confirmed the size range previously observed by TEM: EVs isolated from T98G, U87MG, and U373MG cells ranged from 50 to 200 nm, while EVs derived from U251MG cells exhibited a broader size distribution, extending up to 350 nm. The concentration and size distribution of EVs derived from GBM cell lines were also analysed by NTA [Figure 1E and Supplementary Figure 1B]. All EVs ranged from 40 to 300 nm, with T98G-, U87MG-, and U251MG-derived EVs showing a larger modal diameter of approximately 120 nm, while U373MG-derived EVs had a smaller modal diameter of 90 nm [Figure 1E]. In addition, particle concentration varied among EVs from the different GBM cell lines, with the highest concentration measured for T98G [Supplementary Figure 1B]. Finally, the presence of the main EV markers, such as CD63, TSG101, and ALIX, was assessed using Western blot analysis [Figure 1F]. All EV preparations expressed the analysed markers. CD63 was also detected at the ultrastructural level using immunogold labelling, as illustrated in Figure 1G.

To evaluate TMZ effects on EV release from GBM cells, we first determined the minimum TMZ concentration required to induce cellular stress while maintaining EV production. In line with the Minimal Information for Studies of Extracellular Vesicles (MISEV) guidelines^[19], it is crucial to prevent the contamination of EV samples with apoptotic bodies. Therefore, the sensitivity of various GBM cell lines (T98G, U87MG, U251MG, and U373MG) to TMZ was evaluated using time- and concentration-dependent analyses with the MTT assay. As shown in Figure 2A-D, cell viability was not affected after 24 h of treatment in all four GBM cell lines, whereas a similar decrease in cell viability was observed after 48 and 72 h. Based on these results, treatment with 200 μM TMZ for 48 h was selected for all four GBM cell lines for further investigation of the effects of TMZ on GBM cells and GBM-derived EVs, as this condition did not promote overt cell death. The results obtained using the MTT assay were further validated by a crystal violet cell viability assay, as reported in Supplementary Figure 2. The impact of TMZ on zonal budding distribution on the cell surface was evaluated using a well-established AFM-based approach from our group^[35]. As shown in Figure 2E-I and Supplementary Figure 1C, TMZ treatment resulted in increased Rq, which correlates with budding abundance^[35], at the perinuclear/cellular edge (PNu/CE) area in U251MG and U373MG cells. By contrast, U87MG cells exhibited reduced budding within the nuclear (Nu) region [Figure 2G], whereas T98G cells displayed decreased budding in the Nu region and increased budding at the PNu/CE [Figure 2F]. The effects of TMZ treatment on the size, concentration, and morphology of isolated EVs from all GBM cell lines were analysed. AFM analysis revealed that TMZ treatment increased the mean size of EVs released by U251MG and U373MG cells [Figure 2J and K], while no variation was observed for U87MG cells. Conversely, TMZ treatment reduced the mean diameter of T98G-derived EVs [Figure 2J and K]. Furthermore, NTA analysis revealed a decrease in particle concentration in both T98G- and U251MG-derived EVs under TMZ treatment, with the opposite trend observed for U87MG- and U373MG-derived EVs [Figure 2L and Supplementary Figure 1D]. Morphologically, cryo-EM imaging [Figures 1C and 2M] showed that EVs from both TMZ-treated and untreated cells contained multilayered vesicles, though the relative percentage increased in EVs from treated cells (about 3% CTRL vs. 8% TMZ-treated cells).

Having determined the distinct morpho-dimensional changes induced by TMZ treatment, the effects of the chemotherapeutic agent on the protein cargo of GBM-derived EVs were analysed using label-free quantitative proteomics [Figure 3A-C]. Proteins from EV-SEC fractions of cell-free, depleted culture medium were used as negative controls (see Supplementary Table 1 “CTRL medium” column). As shown in the Venn diagrams of Figure 3A, a total of 84 and 86 proteins were robustly quantified for T98G-derived EVs from TMZ-treated and untreated cells, respectively. For U251MG-derived EVs, 56 proteins were identified in the TMZ-treated condition and 53 in the untreated condition. In U373MG-derived EVs, 162 proteins were quantified following TMZ treatment, compared with 42 proteins in the untreated condition. Lastly, 176 proteins were quantified in EVs released by TMZ-treated U87MG cells and 68 in the untreated condition. The details of the entire quantified protein dataset for each sample are reported in Supplementary Table 1, with protein abundances expressed as iBAQ values. Figures 3A and B demonstrate that TMZ treatment did not significantly impact the protein cargo of T98G-derived EVs, which shared 66.3% of proteins with EVs from untreated cells. These data were confirmed by univariate statistical analysis, as illustrated by the volcano plot [Figure 3B], which showed only minor changes in the EV protein cargo of T98G-derived EVs between the two conditions. In contrast, EVs isolated from TMZ-treated U87MG, U373MG, and U251MG cells exhibited marked alterations in their protein cargo. Specifically, U87MG- and U373MG-derived EVs contained substantially more quantified proteins than EVs from untreated cells, while volcano plot analysis revealed that the majority of differentially expressed proteins were significantly upregulated following TMZ treatment (red dots in Figure 3B). Similarly, U251MG-derived EVs were packaged with a specific set of proteins that were significantly upregulated after TMZ treatment. The complete list of differentially expressed proteins is reported in Supplementary Table 1.

To further investigate the biological effects mediated by EV proteins derived from TMZ-treated GBM cells, the protein ratios (TMZ-treated/untreated) for each GBM cell line were analysed using the IPA tool (accessed in February 2025) to perform GO enrichment analysis, which highlights the significantly modulated biological functions and molecular pathways in which proteins identified in GBM-derived EVs were involved. Figure 3C presents the main significant biological functions and processes as a heatmap, with z-score values shown in red and blue to indicate significant activation and inhibition, respectively, for each comparison between EVs isolated from TMZ-treated and untreated cells. In this context, the functions related to cell migration, adhesion, invasiveness, proliferation, and immune modulation, all processes necessary for tumour progression, were significantly higher in EVs derived from TMZ-treated U87MG and U373MG cells (see the most intense red colouring, which is related to higher and more positive z-score values). In contrast, cell death was significantly reduced in the same comparisons (see the most intense blue colouring related to lower and more negative z-score values). In U251MG cells, the function related to cell migration was also significantly modulated, but to a lesser extent compared with U87MG and U373MG cells. No significant modulation of functions related to proliferation, invasiveness, or immune response was observed in EVs isolated from T98G and U251MG cells, as indicated by the lighter colours of the heatmap boxes. Supplementary Table 1, sheet “downstream”, reports the complete list of downstream effects for each comparison performed with z-score values (TMZ-treated vs. untreated).

In line with the MTT assays and proteomics results, the U87MG cell line (which showed the greatest sensitivity to TMZ treatment) and the U251MG cell line (which showed the lowest sensitivity to TMZ treatment) were selected for all subsequent functional analyses. Furthermore, the proteins involved in cell death processes were downregulated by TMZ treatment in U251MG-derived EVs, whereas they remained almost identical to those in EVs derived from untreated U87MG cells. For this reason, all subsequent functional investigations, including migratory activity, cell death, and immune maturation/activation, were conducted using these two cell lines and EVs derived from them. To this end, EVs isolated from U87MG and U251MG cells, either untreated or treated with TMZ, were used to treat their respective parent cell lines. To further assess potential cross-cell line effects, U87MG cells were incubated with U251MG-derived EVs and U251MG cells with U87MG-derived EVs (see scheme in Figure 3D).

It is important to confirm the internalisation of EVs and assess their uptake rate before investigating their effects on recipient cells. Following the procedure outlined in Figure 3D, U87MG and U251MG cells were incubated with PKH26-labelled GBM-derived EVs for 1, 3, and 24 h. Internalisation was evaluated using fluorescence microscopy. All experimental groups generally showed maximum uptake 24 h after treatment. EVs from untreated GBM cells showed the highest internalization compared with those from TMZ-treated cells [Figure 3E and F, Supplementary Figure 3]. In line with the proteomic analysis, we evaluated GBM cell migratory activity using a wound-healing assay to assess their functional response to EV treatments. As shown in Figure 3C, the cell migration cluster was found to be significantly activated based on the protein cargo of EVs derived from U87MG and U251MG cells after TMZ treatment compared with untreated EVs (z-score values were 6.833 and 2.862, respectively, for U87MG- and U251MG-derived EVs). In addition, 116/176 proteins in U87MG-derived EVs were involved in the cell migration function and were significantly modulated in EVs isolated from TMZ-treated cells, compared to 47/56 proteins in U251MG-derived EVs under the same treatment conditions. As shown in the mechanistic networks of Figure 4A, the majority of these proteins were quantified as unique proteins in EVs derived from U87MG cells treated with TMZ (shown as red shapes in the functional network). Figures 4B and C show the migratory activity of donor GBM cells. TMZ treatment alone did not affect the migratory activity of U87MG cells, but it slowed the migratory activity of U251MG cells over longer periods [Figure 4B and C].

Treatment with EVs derived from U87MG cells, whether exposed to TMZ or not, promoted the migratory activity of recipient U87MG cells as early as 3 h after the scratch was made [Figure 4D (top panel)]. In contrast, only EVs isolated from TMZ-treated U87MG cells enhanced migration in recipient U251MG cells at later time points, with a significant effect observed at 48 h post-scratch [Figure 4D (bottom panel)]. When U87MG cells were treated with EVs derived from both untreated and TMZ-treated U251MG cells, reduced gap closure was observed as early as 1 h post-scratch, indicating reduced migratory activity [Figure 4E (top panel)]. Conversely, U251MG-derived EVs stimulated migration in recipient U251MG cells, but only at later time points, becoming evident at 48 h [Figure 4E (bottom panel)].

The proteomic analysis of EVs revealed that “Organismal Death” was significantly inhibited based on the EV protein cargo of both GBM cell lines when they were treated with TMZ (z-score equal to -7.218 and -3.33 for U87MG- and U251MG-derived EVs, respectively). In particular, as reported in the networks of Figure 5A, U87MG-derived EVs were packaged with proteins that were mainly activated during TMZ treatment, in contrast to U251MG-derived EVs (see red shapes). Western blot analysis revealed an increased expression of key cell death-related markers, including pro-caspase-3, -8, and -9, GPX4 (ferroptosis marker), and microtubule-associated protein 1 LC3B (autophagic marker) in U87MG cells following TMZ treatment [Figure 5B-G]. Conversely, in U251MG cells treated with TMZ, a decrease in the expression of the same markers was observed [Figure 5B-G].

When U87MG cells were treated with EVs isolated from either U87MG or U251MG cells, regardless of whether the parental cells were treated, levels of pro-caspase-3, -8, -9, and GPX4 decreased [Figure 5H-J]. A significant upregulation of these markers was detected in U251MG cells following exposure to EVs isolated from untreated U87MG cells [Figure 5H (bottom panel)]. In contrast, when U251MG cells were treated with EVs derived from U251MG cells, regardless of TMZ treatment, no significant changes were detected in the levels of pro-caspase-3, -8, -9, or GPX4, with the only exception of an increase in pro-caspase-3 levels following incubation with EVs derived from TMZ-treated U251MG cells [Figure 5I (bottom panel) and J]. The evidence shown demonstrates that GBM cells release diverse populations of EVs, whose characteristics and effects are influenced by TMZ treatment.

Proteomics results highlighted a consistent set of EV-associated proteins involved in strong immune activation following TMZ treatment [Figure 3C]. The functional networks shown in Figure 6A predicted significant activation of the “Inflammatory Response” (z-score = 3.825) and “Immune Response of Cells” (z-score = 2.077) biofunctions in U87MG-derived EVs isolated from TMZ-treated cells; in contrast, the protein cargo of U251MG-derived EVs in TMZ-treated condition predicted only a slight, non-significant activation of the “Immune Response of Cells” biofunction (z-score = 1.335). In this regard, the effects of GBM-derived EVs on the activation of BV-2 microglial cells and BMDCs were investigated. These antigen-presenting cells are components of the TME and play a crucial role in the immune response against GBM. For BV-2 activation and polarization, no effects on CD80 and CD206 protein expression were observed after exposure to GBM-derived EVs [Figure 6B and Supplementary Figure 6]. Conversely, analysis of the transcript levels of the pro-inflammatory cytokine IL-1β [Figure 6C] showed a significant increase after treatment with EVs from both untreated and TMZ-treated GBM cells. The polarization of BV-2 cells appeared to be partial, as the expression of IL-1β increased, whereas no upregulation of CD80 was observed. No differences were determined in the mRNA levels of the anti-inflammatory cytokine IL-10 [Figure 6D].

As shown in Figure 6E, PKH26-labelled EVs from both untreated and TMZ-treated U87MG and U251MG cells were rapidly internalised by BMDCs just 30 min after exposure. Specific maturation/activation markers for BMDCs^[40] (CD80, CD86, and MHCII) were evaluated using flow cytometry [Figure 6F-H and Supplementary Figure 7]. Exposure to EVs from both TMZ-treated U87MG and U251MG cells increased expression of CD80, CD86, and MHCII markers in BMDCs compared to EVs derived from untreated GBM cells [Figure 6F-H and Supplementary Figure 7].