Fresh glioblastoma tissue was obtained from a surgical specimen of a 70-year-old female patient treated at the Department of Neurosurgery of our institution in 2022. Integrated diagnosis based on the 2021 fifth edition of the WHO Classification of Central Nervous System Tumors confirmed an IDH1-wildtype glioblastoma. The tissue was registered in our institutional glioma biobank as sample No. 142 and designated as SHG142. Specimen collection was approved in advance by the Ethics Committee of Soochow University (Approval No. SUDA20221206H03, 2022), and the expanded scope of the current project, including organoid establishment and related procedures, was approved under protocol code SUDA20240913H04 (September 2024). All procedures involving human tissue adhered to the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.

Fresh tumor tissue was rinsed twice in phosphate-buffered saline (PBS), minced with sterile scissors, and enzymatically digested with 0.05% trypsin-EDTA (0.2 mM EDTA) at 37 °C for 20–30 min. Dissociated cells were resuspended in complete medium (DMEM/F12 supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin) and seeded into 25 cm² flasks. Cultures were maintained at 37 °C in a humidified 5% CO₂ incubator (Thermo Fisher Scientific, Model 3110), with the medium replenished every 48 h. Upon reaching 80% confluency, the cells were subcultured at a split ratio of 1:3–1:4 using trypsin-EDTA. The cells that maintained stable passage were preserved in freezing medium containing 10% DMSO, initially stored at –80 °C in a step-down freezing container and later transferred to long-term storage. Cells at passage 50 were used for downstream characterization. STR profiling of 21 loci and the amelogenin gene was performed using the SiFaSTR™ 23 Plex Kit to authenticate cell line identity and exclude cross-contamination. Bacterial and mycoplasma contamination were excluded. Manufacturer/supplier information (including catalog numbers) for all commercial reagents, antibodies, and software used in this study is summarized in Supplementary Reagents & Software.

SHG142 cells were seeded in ultralow attachment plates in serum-free GSC medium (DMEM/F12 supplemented with 2% B27 supplement, 20 ng/mL recombinant human epidermal growth factor (EGF), 20 ng/mL basic fibroblast growth factor (bFGF), and 2 µg/mL heparin) supplemented with 1% penicillin‒streptomycin (PS). The cultures were maintained at 37 °C/5% CO₂, with half of the medium changed every 2–3 days. After the formation of tumor spheres, the cells were dissociated into single-cell suspensions. CD133+ cells were isolated using anti-CD133 microbeads (Miltenyi Biotec) according to the manufacturer’s protocol, followed by magnetic-activated cell sorting (MACS). Sorted CD133+ GSCs were expanded in stem cell medium. For passaging, GSCs were centrifuged, washed with PBS, resuspended in fresh medium, and transferred to new flasks. Stem-like properties were validated by immunofluorescence staining of known stem cell markers and in vivo tumorigenicity assays.

For the SHG142 cells, sterile circular coverslips were placed in 24-well plates and seeded with single-cell suspensions. After 24 h of incubation under standard conditions (e.g., 37 °C, 5% CO₂), the cells were fixed in 4% paraformaldehyde (PFA) for 20 min at room temperature. The cells were subsequently washed three times with IF-PBST and permeabilized with 0.5% Triton X-100 for 30 min. After additional washing, the cells were blocked with 5% bovine serum albumin (BSA) for 30 min and incubated overnight at 4 °C with primary antibodies and then with fluorophore-conjugated secondary antibodies for 1 h at room temperature in the dark. After nuclear staining with 4^′,6-diamidino-2-phenylindole (DAPI) (1 µg/mL, 10 min), the cells were mounted using anti-fade mounting medium and imaged with an Olympus fluorescence microscope.

For the SHG142 GSCs, suspended cells were centrifuged at 1000 rpm for 5 min, washed with PBS, fixed with 4% PFA for 20 min, and smeared on slides (20 µL/slide). After air drying, the cells were permeabilized with 0.5% Triton X-100, blocked with 5% BSA, and incubated with primary antibodies overnight at 4 °C. After the samples were washed, secondary antibodies were applied for 1 h in the dark. DAPI staining and imaging were performed as described above. For SHG142 cells, the primary antibodies used were anti-Nestin (Proteintech, Cat. No. 19483-1-AP), anti-GFAP (Proteintech, Cat. No. 16825-1-AP), and anti-Ki-67 (Abcam, Cat. No. ab15580). For SHG142 GSCs, the primary antibodies used were anti-Nestin (Proteintech, Cat. No. 19483-1-AP), anti-CD133 (Proteintech, Cat. No. 18470-1-AP), and anti-A2B5 (Abcam, Cat. No. ab53521). Fluorophore-conjugated secondary antibodies were purchased from Invitrogen (Cat. Nos. A-11008 and A-11005).

Exponentially growing SHG142 cells were seeded into 96-well plates at 1 $\times$ 10³ and 2 $\times$ 10³ cells/well in a 100-µL volume. Peripheral wells were filled with PBS to reduce edge effects. At 0, 24, 48, 72, 96, and 120 h, 10 µL of CCK-8 reagent was added to each well. The plates were incubated for 2 h at 37 °C before the absorbances were measured at 450 nm using a microplate reader.

Cells in the logarithmic growth phase were treated with colchicine (0.25 µg/mL) for 3 h to arrest the cells in metaphase. After enzymatic dissociation and centrifugation, the cells were subjected to hypotonic treatment with 75 mM KCl at 37 °C for 15 min, followed by two rounds of fixation in methanol: acetic acid (3:1). The suspension was dropped onto precooled slides and incubated at 80 °C. Chromosomes were stained with Giemsa and examined under a microscope. Karyotypes were classified according to the International System for Human Cytogenetic Nomenclature (ISCN).

Twelve BALB/c nude mice were purchased from the Model Animal Research Center of Nanjing University and housed in the SPF animal facility at our institution. The mice had free access to autoclaved water and standard rodent chow and were acclimatized for at least one week prior to experimentation. Ethical approval for this study was obtained from the Ethics Committee of Soochow University (Approval No. SUDA20240913A03).

Luciferase-expressing SHG142 cells or SHG142 GSCs were resuspended in PBS (4 $\times$ 10⁴ cells/µL). Under isoflurane anesthesia (5% induction, 1–2% maintenance), the mice (n = 6/group) were secured in a stereotactic frame. Isoflurane was delivered continuously via a calibrated vaporizer and facemask, and an adequate surgical plane of anesthesia was confirmed by the absence of the pedal withdrawal reflex (toe pinch). The animals were continuously monitored throughout the procedure to ensure appropriate anesthesia depth and animal welfare. A midline scalp incision was made, and a burr hole was drilled 2 mm anterior to the bregma and 3.5 mm lateral to the sagittal suture. Cells (5 µL) were injected into the right caudate nucleus (3 mm depth) at 1 µL/min using a Hamilton syringe. The needle was withdrawn slowly, the burr hole was sealed with bone wax, and the incision was sutured. Tumor growth was monitored by in vivo bioluminescence imaging (IVIS) on Days 7 and 14. Survival was defined as the time (in days) from intracranial implantation (Day 0) to the humane endpoint. Humane endpoints were predefined, and the animals were euthanized immediately when they developed overt neurological deficits or when tumor progression exceeded the predefined endpoint threshold (estimated tumor burden $>$10%), as assessed by clinical evaluation together with in vivo bioluminescence imaging. Euthanasia was performed by intraperitoneal administration of an overdose of pentobarbital sodium (150 mg/kg, 50 mg/mL). Death was confirmed by cessation of respiration, absence of a heartbeat, and loss of reflexes, in accordance with the AVMA Guidelines for the Euthanasia of Animals. The brains were harvested for H&E and IHC analyses. Kaplan–Meier curves were used for survival analysis.

Statistical analysis of survival data was performed using GraphPad Prism (version 10.1.2; GraphPad Software, San Diego, CA, USA). Kaplan–Meier survival curves were generated for mice implanted with SHG142 cells or SHG142 GSCs (n = 6 per group) and compared using the log-rank (Mantel–Cox) test. Exact p values are reported, and a p value 0.05 was considered to indicate statistical significance.

Stable luciferase-expressing cells were generated using a lentiviral transfer vector GV281 (GeneChem, China), in which firefly luciferase is driven by the human Ubc promoter (Ubc–firefly luciferase–MCS–IRES–puromycin). Stable GFP-expressing cells were generated using a lentiviral transfer vector GV493 (GeneChem, China), in which EmGFP is driven by the CBh promoter (hU6–MCS–CBh–EmGFP–IRES–puromycin). Cells were transduced with lentiviral particles and subsequently selected with puromycin to establish stable populations (selection conditions were optimized by kill-curve testing).

Human cerebral organoids were generated from H9 embryonic stem cells (hESCs) obtained from the National Collection of Authenticated Cell Cultures (China). Organoid differentiation and maturation were performed with reference to established cerebral organoid protocols [14, 15]. Briefly, 9000 hESCs/well were seeded in 96-well ultralow attachment plates in STEMdiff™ Neural Induction Medium to form embryoid bodies (EBs). EBs were transferred to neural induction medium for neuroectoderm specification. Neuroepithelial tissues were embedded in Matrigel droplets and cultured in IDM-A medium in 24-well plates on an orbital shaker for expansion. IDM-A medium was prepared by mixing 125 mL DMEM/F12 and 125 mL Neurobasal medium (Gibco, 21103-049), supplemented with 1.25 mL N-2 (Gibco, 17502-048), 62.5 µL insulin (Sigma-Aldrich, I9278), 2.5 mL GlutaMAX (Gibco, 35050-061), 1.25 mL MEM-NEAA (Gibco, 11140-050), 2.5 mL penicillin–streptomycin (Gibco, 15140-122), and 5 mL B-27 minus vitamin A (Gibco, 12587-010).

On Day 30 of organoid maturation, 1 $\times$ 10⁴ SHG142-GSCs were cocultured with individual organoids. After 24 h of static culture, the organoids were gently washed with PBS. Cocultures were maintained for 7 days, fixed in 4% PFA, and analyzed by IF staining for tumor infiltration and marker expression (e.g., Ki-67, CD133, Nestin, and SRY-Box Transcription Factor 2 (SOX2)).

To evaluate drug response in a more physiologically relevant 3D context, SHG142 GSCs stably expressing luciferase were cocultured with brain organoids to establish SHG142 organoid-based coculture tumor models. After coculture maturation, organoid cultures were treated with vehicle control or single-dose drug exposures at reference concentrations informed by the in vitro IC50 values: temozolomide (TMZ, 900 µM), lomustine (CCNU, 120 µM), carmustine (BCNU, 60 µM), or SN-38 (20 nM). Bioluminescence was assessed at Day 0 (immediately prior to drug treatment) and at Day 7 post-treatment using plate-based luminescence imaging, and the treatment effect was quantified as luciferase fold change (Day 7/Day 0) for each organoid. Each condition included n = 3 organoids. Statistical analyses were performed in GraphPad Prism using ordinary one-way ANOVA followed by Dunnett’s multiple comparisons test versus the vehicle control; adjusted p values were reported.

Genomic DNA was extracted from parental bulk primary glioma cells (P2) and P50 SHG142 cells and subjected to quality control. The qualified DNA was fragmented into 150–350 bp fragments. Libraries were constructed via end repair, A-tailing, and adapter ligation. Libraries were hybridized with biotinylated exon capture probes (Illumina) and enriched with streptavidin-coated beads. The captured DNA was amplified via PCR. The final libraries were quantified and sequenced on an Illumina NovaSeq 6000 platform (PE150 configuration). Variant allele frequency (VAF) was calculated as Alt/(Ref+Alt) from read counts, and variants were annotated and summarized for concordance analysis between P2 and P50. For the SNV-only concordance analysis, variants were filtered using harmonized criteria in both datasets: exonic variants were retained and the TERT promoter hotspot–region variant annotated as an upstream/promoter event was included; synonymous SNVs were excluded; only PASS variants were kept; and common polymorphisms were removed using gnomAD (allele frequency 0.01).

Primary tumor tissue was dissociated into single cells and cultured in serum-containing medium. The tumor cells proliferated slowly during the initial five passages, after which their growth rate increased and stabilized. SHG142 cells formed a monolayer and displayed an irregular morphology, with most cells appearing bipolar or spindle shaped (Fig. 1A). The nuclei showed marked pleomorphism, ranging from round to irregular polygonal shapes, and individual cells often contained one or two nuclei. A small proportion of cells exhibited pseudopod-like extensions, whereas most cells lacked prominent projections. These morphological features remained stable during subsequent passages. Immunofluorescence staining of passage 50 cells revealed positive expressions of Nestin, GFAP and Ki-67 (Fig. 1H).

Fig. 1.

Morphological and phenotypic characterization of the SHG142 glioblastoma cell line and its glioma stem-like cells. (A) Bright-field image showing the morphology of SHG142 cells at passage 50 (scale bar: 200 µm). (B) Morphology of SHG142 GSC spheres at passage 1 (scale bar: 200 µm). (C) Morphology of SHG142 GSC spheres at passage 5 (scale bar: 200 µm). (D) Morphological changes of SHG142 GSCs after 6 h of adherent culture in DMEM/F12 supplemented with 10% fetal bovine serum but without EGF or bFGF (scale bar: 200 µm). (E) As in (D) after 24 h (scale bar: 200 µm). (F) As in (D) after 48 h (scale bar: 200 µm). (G) Cell proliferation curve of SHG142 cells; the x-axis indicates time, and the y-axis shows the absorbance at 450 nm. (H) Immunofluorescence staining showing the expressions of key markers in passage 50 SHG142 cells: Nestin (++), GFAP (++), and Ki-67 (50%) (scale bar: 50 µm). (I) Representative immunofluorescence images of CD133, Nestin, and A2B5 expressions in tumor spheres formed by SHG142 GSCs. DAPI was used as a nuclear counterstain for immunofluorescence (scale bar: 50 µm). (J) STR genotyping profile of SHG142 cells at passage 50. Abbreviations: A2B5, Ganglioside Antigen A2B5; DAPI, 4^′,6-diamidino-2-phenylindole; GFAP, glial fibrillary acidic protein; GSC, glioma stem-like cell; STR, short tandem repeat; EGF, epidermal growth factor; bFGF, basic fibroblast growth factor.

Passage-50 SHG142 cells were enzymatically dissociated, centrifuged, resuspended in neural stem cell medium (NSCM) and seeded into ultralow-attachment plates for suspension culture. After 5–7 days, numerous floating spheroids with diameters greater than 50 µm were observed and designated as SHG142 GSCs (Fig. 1B,C). During passaging, only the suspended spheroids were collected, and any adherent cells were discarded. Immunofluorescence staining of SHG142 GSCs at passage 5 confirmed the positive expressions of the glioma stem-like cell markers, CD133, Nestin and A2B5 (Fig. 1I). To assess their differentiation capacity, SHG142 GSCs were transferred to DMEM/F12 supplemented with 10% FBS but lacking bFGF and EGF. After 6 h, partial cell attachment was observed, with adherent cells accumulating around the original spheroids; by 48 h, nearly all the cells had adhered to the substrate (Fig. 1D–F).

STR profiling at passage 50 confirmed that SHG142 is a female-derived glioma cell line. The STR pattern did not match that of any existing cell line in the ExPASy STR database, indicating that SHG142 represents a unique cell line (Fig. 1J).

The proliferative capacity of passage 50 SHG142 cells was evaluated using a Cell Counting Kit-8 (CCK-8) assay (Fig. 1G). The optical density (OD) at 450 nm increased modestly over the first 48 hours, followed by a marked increase thereafter. The growth curve of the SHG142 cells exhibited a biphasic pattern: an initial lag phase with limited proliferation within the first 48 hours, followed by a transition into a logarithmic growth phase, ultimately reaching a plateau indicating saturation. For limited benchmarking against a widely used GBM reference line (LN229) under identical conditions, a head-to-head CCK-8 comparison is provided in Supplementary Fig. 1.

The SHG142 glioblastoma cell line exhibited pronounced chromosomal instability (CIN), a hallmark of malignancy characterized by numerical abnormalities (e.g., aneuploidy) and large-scale structural alterations, such as translocations, deletions, and amplifications. On the basis of previous reports, the chromosomal numerical abnormalities in some glioma cell lines are summarized in Table 1 (Ref. [5, 11, 16, 17, 18]).

G-banding karyotype analysis was performed on passage 50 SHG142 cells, with 100 metaphase spreads analyzed. The chromosome numbers ranged from 36 to 94, approximately forming two distinct populations: nearly half of the cells maintained a near-diploid state with 44–47 chromosomes, whereas approximately 20% displayed hyperdiploid or polyploid profiles (Fig. 2B).

Fig. 2.

Karyotypic analysis of SHG142 glioblastoma cells. (A) Representative metaphase spread of SHG142 cells from a single-karyotype analysis. (B) Distribution of SHG142 cells by chromosomal ploidy category (n = 100). (C) Corresponding karyogram for the metaphase shown in (A).

Clonal numerical aberrations, including trisomy of chromosome 7 and monosomy of chromosome 10, were recurrently observed. Monosomies of chromosomes 13 and 17 were also frequently detected. Structural rearrangements included a breakpoint at 3p13 with a translocation to an unidentified chromosome, a t(9q34;13q22) translocation, a t(13q12;3p12) translocation, and a deletion within 11q14 (Fig. 2A,C). These findings are consistent with the known genomic complexity of high-grade gliomas.

Preoperative magnetic resonance imaging (MRI) revealed a mass in the right temporal lobe. On T1-weighted imaging (T1WI), the lesion exhibited slightly hypointense signals (Fig. 3A), whereas contrast-enhanced scans demonstrated an irregular ring-like enhancement with a nonenhancing central region suggestive of necrosis (Fig. 3B). T2-weighted imaging (T2WI) revealed a mildly hyperintense signal in the solid portion of the tumor, surrounded by a large area of hyperintense peritumoral edema, with evidence of compression of the lateral ventricle (Fig. 3C).

Fig. 3.

MRI, hematoxylin and eosin (H&E) staining, and immunohistochemical of tumor tissues from the patient and intracranial xenografts. (A) Preoperative T1-weighted MRI. (B) Preoperative contrast-enhanced T1-weighted MRI. (C) Preoperative T2-weighted MRI (scale bar: 5 cm). (D) H&E staining of the patient’s primary tumor (scale bar: 100 µm). (E) H&E staining of intracranial xenograft tumor derived from SHG142 cells (scale bar: 2000 µm; inset: 100 µm). (F) H&E staining of intracranial xenograft tumor derived from SHG142 GSCs (scale bar: 2000 µm; inset: 100 µm). (G) Comparative immunohistochemical staining of GFAP, Nestin, MGMT, IDH1, Ki-67 (~50%), P53 (~80%), S-100 and CD34 between the patient’s primary glioblastoma tissue and xenograft tumor tissue in nude mice (scale bar: 50 µm). MRI, magnetic resonance imaging; MGMT, O6-methylguanine-DNA methyltransferase; IDH1, isocitrate dehydrogenase 1.

Histopathological examination of the resected tissue revealed a densely cellular tumor composed of pleomorphic and atypical cells. The tumor nuclei were enlarged, hyperchromatic, and irregular in shape, with frequent mitotic figures observed. The tumor displayed an infiltrative growth pattern with a lack of normal tissue architecture. Neovascularization foci were also noted (Fig. 3D).

Immunohistochemical staining revealed positive expressions of GFAP, Nestin, MGMT, Ki-67, P53, S-100 and CD34, while IDH1 staining was negative. The Ki-67 proliferation index was approximately 50%, indicating a high proliferative capacity of the tumor cells (Fig. 3G).

To evaluate the tumorigenicity of SHG142 and SHG142 GSCs in vivo, we orthotopically implanted each cell population into the brains of immunodeficient mice. Both SHG142 and SHG142 GSCs formed intracranial tumors (Fig. 4A), with SHG142 GSCs demonstrating modestly greater tumorigenic behavior, as evidenced by a shorter median survival in mice bearing SHG142 GSC-derived tumors than in those implanted with SHG142 cells (median survival: 25 vs. 31.5 days, p = 0.025; log-rank test; HR = 3.179; 95% CI 0.845–11.96) (Fig. 4B). Given the limited cohort size (n = 6 per group), this survival difference should be interpreted cautiously.

Fig. 4.

Tumorigenic potential of the SHG142 cells and SHG142 GSCs in intracranial xenografts. (A) Bioluminescence imaging of mouse brains at Day 7 and 14 after intracranial implantation of SHG142 cells and SHG142 GSCs. (B) Kaplan–Meier survival curves of mice implanted with SHG142 cells and SHG142 GSCs (n = 6 per group); p = 0.025, log-rank test; HR = 3.179, 95% CI 0.845–11.96. Survival was defined as the time from intracranial implantation to the humane endpoint. (C–H) Immunohistochemical staining of intracranial xenograft tumors derived from SHG142 GSCs, showing positive expressions of (C) Nestin, (D) CD133, (E) A2B5, (F) SOX2, (G) vimentin, and (H) CD34 (scale bar: 50 µm).

Histological analysis of intracranial xenografts derived from SHG142 cells revealed well-defined tumor boundaries and localized growth. The tumors displayed high cell densities, with predominantly spindle- and round-shaped cells (Fig. 3E). Immunohistochemistry demonstrated positive expressions of GFAP, Nestin, MGMT, Ki-67, P53, S-100 and CD34, with a Ki-67 labeling index of approximately 50%, and IDH1 expression was not detected (Fig. 3G). These results are consistent with those observed in the patient’s primary tumor tissue.

In contrast, tumors derived from SHG142 GSCs exhibited a more infiltrative growth pattern with an ill-defined tumor margin and areas of tumor–brain interface disruption. These tumors were highly cellular and displayed greater cytomorphologic heterogeneity, accompanied by prominent tissue clefting near the tumor–brain interface and focal hemorrhage with erythrocyte extravasation, suggestive of a more invasive phenotype (Fig. 3F). Immunohistochemical staining confirmed the positive expressions of stemness and mesenchymal markers, including Nestin, CD133, A2B5, SOX2, vimentin, and CD34 (Fig. 4C–H).

In addition to in vivo models, brain organoids were utilized to simulate the invasion and tumor formation of SHG142 GSCs in a human-relevant 3D environment. H9 human embryonic stem cells (hESCs) were first thawed and cultured. On day 7, the formation of EBs was induced in the hESCs. By day 14, neural induction medium was applied to initiate neuroectodermal differentiation. On day 21, the resulting neuroepithelial tissue was transferred onto Matrigel droplets to promote neuroepithelial expansion. Matrigel-embedded tissues were subsequently cultured in 24-well plates on an orbital shaker to facilitate cerebral tissue growth and expansion. Mature brain organoids were formed after approximately 50 days (Fig. 5A). Stable GFP-expressing SHG142 GSCs were first established via lentiviral transduction to enable visualization and tracking within the organoid system (Fig. 5B). Immunofluorescence analysis on Day 7 after coculture revealed early infiltration and tumorigenic behavior of SHG142 GSCs within the cerebral organoids (Fig. 5C), and the invasion phenotype was further quantified, as shown in Supplementary Fig. 2 (invasion-positive rate, invasion depth, and % infiltrated area). Prior to tumor cell seeding, the developmental fidelity of the hESC-derived cerebral organoids was validated through assessments of tissue architecture, differentiation progression, and cellular composition (Fig. 5D–F). After mature organoids were cocultured with SHG142 GSCs for 7 days, immunofluorescence analysis demonstrated robust expressions of Ki-67, CD133, Nestin, and SOX2, confirming tumor cell proliferation and the retention of stem-like properties within the organoid system (Fig. 5G).

Fig. 5.

Generation of H9 hESC-derived brain organoids and application to SHG142 GSC co-culture and drug response assessment. (A) Representative bright-field images illustrating the organoid differentiation timeline: H9 ESC culture (Day 7) $\to$ Embryoid body (EB) formation (Day 14) $\to$ Expanded neuroepithelium (Day 21) $\to$ Mature cerebral tissue (Day 50) (n = 4; scale bar: 200 µm). (B) Stable GFP-expressing SHG142 GSC established via lentiviral transduction (scale bar: 200 µm). (C) Immunofluorescence analysis of SHG142 GSCs infiltration and tumor formation within human brain organoids on Day 7 after coculture (n = 4; scale bar: 1000 µm). (D) Immunofluorescence showing differentiation from Musashi-1-positive neural stem/progenitor cells to NeuN-positive mature neurons on day 50; nuclei were counterstained with DAPI (scale bar: 50 µm). (E) PAX6-positive dorsal forebrain neural stem cells aligned along an N-cadherin-marked apical membrane (scale bar: 50 µm). (F) Nestin+/GFAP+ double-positive cells represent the intermediate state of radial glia transitioning into astrocytes (scale bar: 50 µm). (G) After 7 days of coculture, the SHG142 GSC brain organoids expressed Ki-67, CD133, Nestin, and SOX2 (n = 4; scale bar: 50 µm). (H) Representative plate-based bioluminescence images at Day 0 (pre-treatment) and Day 7 after single-dose exposure to vehicle, TMZ (900 µM), CCNU (120 µM), BCNU (60 µM), or SN-38 (20 nM); pseudocolor indicates radiance (p/sec/cm²/sr). (I) Quantification of therapeutic response expressed as luciferase fold change (Day 7/Day 0). Each dot represents one organoid (n = 3 per group); bars indicate mean $\pm$ SD. Statistical significance was assessed by ordinary one-way ANOVA followed by Dunnett’s multiple comparisons test versus vehicle (adjusted p 0.0001 for all comparisons). In vitro IC50-based dose–response profiling and benchmarking versus LN229 are provided in Supplementary Fig. 3. Abbreviations: hESC, human embryonic stem cell; GFP, green fluorescent protein; BCNU, carmustine; CCNU, lomustine; SN-38, active metabolite of irinotecan; TMZ, temozolomide.

To complement the organoid invasion readouts, we further performed single-dose drug testing in the SHG142 GSC–organoid coculture model with plate-based bioluminescence imaging. Representative bioluminescence images at Day 0 and Day 7 show robust tumor-associated signal expansion in the vehicle group and marked suppression following drug exposure (Fig. 5H). Quantification using baseline-normalized luciferase fold change (Day 7/Day 0) demonstrated significant growth suppression across all four agents compared with vehicle (vehicle mean fold change 4.103 vs. TMZ 1.647, CCNU 1.590, BCNU 0.953, and SN-38 0.843; n = 3 organoids per group; ordinary one-way ANOVA followed by Dunnett’s multiple comparisons test vs. vehicle, adjusted p 0.0001 for all comparisons) (Fig. 5I). Reference in vitro IC50 values and the full dose–response comparisons versus LN229 are provided in Supplementary Fig. 3.

Genetic mutations are central driving forces in tumorigenesis. In glioblastoma (GBM), oncogenesis is closely associated with recurrent driver alterations, such as EGFR amplification, TERT promoter mutations, TP53 inactivating mutations, PTEN deletions, and IDH1 R132H somatic mutations. These alterations synergistically contribute to malignant glioma clonal evolution by disrupting key signaling pathways, including the RTK/PI3K/AKT axis, the p53-mediated genomic stability machinery, and the RB-associated cell cycle regulatory network. TERT promoter mutations, which create de novo binding motifs for ETS transcription factors (notably GABP), lead to aberrant reactivation of telomerase in tumor cells, enabling replicative immortality—a hallmark of glioblastoma pathogenesis.

To assess genomic concordance and lineage continuity during in vitro propagation, we performed WES on the parental early-passage bulk primary culture (P2), derived from the same surgical specimen, in parallel with SHG142 cells at passage 50. The results revealed that key glioblastoma-associated events, such as TP53 p.H193R, truncating NF1 p.R1947X, and the TERT promoter hotspot (c.-124C$>$T), were retained from P2 to P50 (Table 2). To enable a harmonized SNV-only comparison, we applied identical filtering criteria to P2 and P50 callsets (exonic variants retained with inclusion of the upstream/promoter TERT hotspot–region variant; synonymous SNVs excluded; PASS variants retained; gnomAD allele frequency 0.01). After filtering, 498 P2 SNVs and 698 P50 SNVs were used for concordance analysis. Across SNV calls, a substantial proportion of variants were shared between P2 and P50 (329 shared SNVs; 66.1% of P2 SNVs), supporting broad mutational concordance between the parental culture and the established SHG142 line (Supplementary Table 1).