Glioma tissues were collected from 20 patients admitted to Wuxi No. 2 People's Hospital (Wuxi, China) for surgical resection between January 2020 and January 2022. The 20 patients included 12 men and 8 women, aged between 26 and 70 years old, with an average age of 50.04±12.34 (mean ± SD) years. The patients did not receive radiotherapy, chemotherapy, targeted therapy or any other treatment that may affect the tumor, and did not have any other tumors or systemic diseases. A total of 16 patients had GBM [World Health Organization (WHO) grade IV] (26) and 4 patients had astrocytoma (WHO grade III). Additionally, 20 tumor-adjacent normal brain tissues (NBTs) were collected from the patients and were used as the control. The present study was approved by the Ethics Committee of Wuxi No. 2 People's Hospital (approval no. 2022-Y-116) and all patients provided written informed consent.

The following human GBM cell lines were used: A172 (cat. no. CL-0012; Wuhan Punosai Life Technology Co., Ltd.), SHG44 (cat. no. BH-0109468; Shanghai Bohu Biotechnology Co., Ltd.), U87 MG (cat. no. FH0162; GBM of unknown origin) and U251 (cat. no. FH0159) (both from FuHeng Biology). A172, SHG44 and U87-MG cells were cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (FBS; cat. no. F0193; Shanghai Noning Biotechnology Co., Ltd.) and 100 U/ml penicillin/streptomycin (cat. no. G4003; Wuhan Servicebio Technology, Co., Ltd.) at 37°C with 5% CO₂. U251 cells were cultured in DMEM/F12 medium (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% FBS and 100 U/ml penicillin/streptomycin at 37°C with 5% CO₂.

To induce differentiation, A172, SHG44, U87 MG and U251 cells were resuspended in neurobasal medium (cat. no. 21103049; Gibco; Thermo Fisher Scientific, Inc.) containing 20 ng/ml bFGF (cat. no. C046), 20 ng/ml EGF (cat. no. C029), 1 mg/ml heparin (cat. no. CK98) (all from Novoprotein Scientific, Inc.) and 100 U/ml penicillin/streptomycin. Cells were adjusted to 2×10⁵/ml for culture and neurobasal medium was replaced every 2-3 days. When the diameter of the glioma tumor spheres reached 150-200 μm (after ~7 days), the glioma tumor spheres were collected and digested in an Accutase solution (cat. no. A6964; MilliporeSigma). A single cell suspension was prepared and subcultured.

The miR-1205 mimics and mimics negative control (NC) were synthesized by Shanghai GenePharma Co., Ltd., and 50 nM mimics were mixed with Lipofectamine^® 2000 (Invitrogen; Thermo Fisher Scientific, Inc.) in OptiMEM (Gibco; Thermo Fisher Scientific, Inc.) to form complexes at room temperature, which were then transfected into the GSC-U87 and GSC-U251 cells at 37°C for a 2-3 day incubation, and RNA and protein extraction were performed after 48 h.

LINC00900/YTHDF1 overexpression/short hairpin RNA (shRNA) and control/shRNA-NC (sh-NC) lentiviruses (Plvx-shRNA2-mcherry-T2A-Puro and pGMLVCMV-M CS-3xflag-EF1-Zsgreen1-T2A-puro) were purchased from OLIGOBIO. To obtain cells with overexpression or knock down of LINC00900 and YTHDF1, GSC-U87 and GSC-U251 cells were infected with lentiviruses (MOI=10) at 37°C for 48 h, and cultured in medium supplemented with 5 μg/ml puromycin after infection. The cells were collected for further experiments at 2 weeks post-infection. The sequences of the short hairpin (sh)RNAs and miRNA mimics are listed in Table I.

Total RNA was extracted from tissues or cells using an RNA extraction kit (cat. no. R1200; Shanghai Yuduo Biotechnology Co., Ltd.) in accordance with the manufacturer's instructions, and the RNA concentration was measured using an ultraviolet spectrophotometer. The extracted RNA was then reverse transcribed into cDNA using a RT kit according to manufacturer's protocol (cat. no. PC1703; Aidlab Biotechnologies Co., Ltd.), followed by qPCR analysis using a SYBR Green PCR kit (cat. no. PC3302; Aidlab Biotechnologies Co., Ltd.). The qPCR thermocycling conditions were as follows: Denaturation at 94°C for 30 sec; followed by 30-35 cycles of denaturation at 94°C for 30 sec, annealing at 55-60°C for 30 sec, and extension at 72°C for 24 sec; and a final extension step at 72°C for 5-10 min. The 2^−ΔΔCq method (27) was used to analyze the relative expression levels of the target gene, and GAPDH was used as endogenous control. Primer sequences are listed in Table II.

Total proteins were extracted from tissues or cells using RIPA buffer (cat. no. P0013B; Beyotime Institute of Biotechnology) in accordance with the manufacturer's instructions. A BCA kit (cat. no. BL521A; Biosharp Life Sciences) was used to measure the protein concentration. The proteins (25 μg) were then separated by SDS-PAGE (Mini-PROTEAN^® 3 cell; Bio-Rad Laboratories, Inc.) on 8% gels and were transferred to a PVDF membrane (cat. no. IPVH00010; MilliporeSigma). Subsequently, 5% dry skim milk was used to block the membranes at room temperature for 3 h. The following primary antibodies were then applied at 4°C overnight: Anti-YTHDF1 (1:1,000; cat. no. ab220162; Abcam), anti-STAT3 (1:1,000; cat. no. ab68153; Abcam), anti-NANOG (1:1,000; cat. no. ab109250; Abcam), anti-OCT4 (1:1,000; cat. no. ab137427; Abcam) and anti-GAPDH (1:5,000; cat. no. 60004-1-Ig; Proteintech Group, Inc.), followed by the secondary antibodies (1:5,000; cat. nos. ZB2301 and ZB2305; OriGene Technologies, Inc.) at room temperature for 2 h. Target protein bands were developed by enhanced chemiluminescence (cat. no. ECL-0011; Beijing Dingguo Changsheng Biotechnology Co., Ltd.) and detected using a chemiluminescence imager (ChemiScope 5300 Pro; Shanghai Huanxi Medical Equipment Co., Ltd.).

Total RNA was extracted from tissues, and the Magna m6A RNA methylation immunoprecipitation kit (cat. no. C11051-1; Guangzhou Ruibo Biotechnology Co., Ltd.) was used for MeRIP in accordance with the manufacturer's instructions. The enriched RNA was then analyzed by RT-qPCR as aforementioned.

The induced GSCs were collected, washed with PBS and fixed with 4% paraformaldehyde for 30 min at 4°C. Then, the anti-CD133 primary antibody (1:500; cat. no. 66666-1-Ig; Proteintech Group, Inc.) was applied at 4°C overnight. Subsequently, the corresponding Cy3-conjugated Affinipure Goat Anti-Mouse IgG (H+L) secondary antibody (1:80; cat. no. SA00009-1; Proteintech Group, Inc.) was applied at 37°C for 1 h and a Hoechst 33342 staining solution (10 μg/ml; cat. no. 875756-97-1; MilliporeSigma) was applied for 10 min at room temperature. CD133 immunofluorescence was observed under a fluorescence microscope (DP70; Olympus Corporation).

Suspensions of cells transfected with miRNA mimics or infected with LINC00926/YTHDF1 overexpression/knockdown lentiviruses were seeded in a 96-well plate at 2,000 cells/well. After 24 h, 10 μl Cell Counting Kit (CCK)-8 solution (cat. no. C0038; Beyotime Institute of Biotechnology) was added to each well for 2 h at 37°C. A microplate reader (MK3; Thermo Fisher Scientific, Inc.) was used to measure absorbance at 450 nm.

GSCs (2×10⁵) were collected and seeded in the upper chamber of a 24-well Transwell system (pore size, 8 μm) precoated with Matrigel (cat. no. 354234; Shanghai Yanhui Biotechnology Co., Ltd.) at 37°C for 1 h, and culture medium containing 10% FBS was added to the lower chamber. Cells were incubated for 24 h at 37°C, and then the medium in the upper chamber was discarded. Cells were fixed with 4% 40 g/l paraformaldehyde for 15 min at 4°C, washed with PBS and then stained with 0.2% crystal violet at room temperature for 5 min. Cells on the bottom membrane of the chamber were gently wiped with a cotton swab. After drying, the chamber was observed under an optical microscope (DFC300 FX; Leica Microsystems, Inc.) and cells were counted.

A total of 2,000 GSCs/well were seeded in the well of a 24-well plate in medium supplemented with 20 mg/ml B27 (cat. no. 17504044; Gibco; Thermo Fisher Scientific, Inc.), 20 ng/ml EGF and 20 ng/ml bFGF. EGF and bFGF were added to the medium twice a week. After 2-3 weeks of culture, the number of spheres formed in each well was observed under an optical microscope. Sphere formation efficiency (SFE) was used to evaluate stem cell self-renewal, as follows: SFE=number of cell spheres with a >75 μm diameter per well/total number of original seeded cells in per well.

GSC-U87 and GSC-U251 cells were lysed using IP lysis solution (cat. no. ZN2923; Beijing Baiao Leibo Technology Co., Ltd.) on ice for 15 min and the supernatant was collected by centrifugation (4°C, 16,000 × g, 10 min). The biotin-labeled LINC00900 probe and negative control probe were synthesized by Guangzhou RiboBio Co., Ltd. Subsequently, 0.4 μg (50 pmol) biotin-labeled RNA was incubated with 0.5 ml cell lysate at 4°C for 1 h. Then, 50 μl protein G agarose beads were added and incubated at 4°C for 1 h. Agarose beads were centrifuged at 13,400 × g, for 30 sec at 4°C for precipitation and the supernatant was collected into a clean tube. After washing, western blotting was performed to detect the target protein as aforementioned.

RIP was conducted using a RIP kit [cat. no. GS-ET-006; Genseq; Hylegen Biotechnology (Shanghai) Co., Ltd.] in accordance with the manufacturer's instructions. Cells were lysed using 250 μl complete lysis buffer for 5 min, and 50 μl lysate was then incubated with IgG (1:100, cat. no. 36111ES10) or an anti-YTHDF1 antibody (1:500; cat. no. 31040ES50) (both from Shanghai Yeasen Biotechnology Co., Ltd.). RNA was extracted and purified, and RT-qPCR was used to measure target gene expression, as aforementioned.

Cells infected with LV-sh-NC and LV-sh-YTHDF1 lentiviruses [Hanheng Biotechnology (Shanghai) Co., Ltd.] were seeded in 6-well plates. After 24 h of culture, the cells reached 70% confluence. The cells were then treated with 5 μg/ml actinomycin D (cat. no. 1036; BioVision; Abcam) for 0, 2 or 4 h at 37°C, and the expression levels of LINC00900 were measured by RT-qPCR as aforementioned.

Prediction analysis (DIANA-LncBase v.2, http://diana.imis.athena-innovation.gr/DianaTools/index.php?r=site/page&view=software; and TargetScan, https://www.targetscan.org/vert_71/) revealed a miR-1205-binding site in the LINC00900 transcript and STAT3 3′UTR. LINC00900 and STAT3 3′-UTR wildtype (WT) and mutant type (MUT) luciferase reporter genes (pmirGLO) were constructed by Beijing Olinger Biotechnology Co., Ltd. and co-transfected with miR-1205 mimics or mimics NC into 293T cells (cat. no. JN15568; Shanghai Jining Biotechnology Co., Ltd) by Lipofectamine 2,000. After 48 h, the dual luciferase reporter gene detection kit (cat. no. E1910; Promega Corporation) was used to detect the firefly luciferase activity and Renilla luciferase activity. The ratio of firefly luciferase to Renilla luciferase was calculated, and the control group was set as 1 to calculate the relative luciferase activity of different treatment groups.

Male nude specific pathogen-free mice (n=20; age, 3-4 weeks) were purchased from Beijing Huafukang Biotechnology Co., Ltd. The mice were maintained at 20-26°C and 50-60% relative humidity. They were supplied with ad libitum access to sufficient food/water and were kept under a 12-light/dark cycle. The nude mice were divided into the following four groups (n=5/group): LV-sh-NC1 + LV-sh-NC2, LV-sh-YTHDF1, LV-sh-LINC00900 and LV-sh-YTHDF1 + LV-sh-LINC00900. All mice were subcutaneously injected with 5×10⁶ GSC-U87 cells infected with LINC00900/YTHDF1 knockdown lentiviruses. After 1 week, the tumor growth curve and diameter were measured to calculate tumor volume (1/2 × length × width²). All experiments were conducted in accordance with protocols approved by the Medical Ethics Committee of Wuxi No. 2 People's Hospital (Wuxi, China, approval no. 2023-Y-200).

If the animals reached any of the following humane endpoints they were sacrificed: The maximum tumor diameter exceeded 15 mm; animal weight loss reached ≤20%; the animal exhibited cachexia or wasting syndrome; the size of the solid tumor exceeded 10% of body weight. The experimental duration of the present study was 4 weeks, with the first week being used for adaptive feeding. At the beginning of the second week, the experiment was initiated (subcutaneous injection of GSCs to establish a xenograft tumor animal model) and animal status was observed daily. Tumor volume was measured from the first week after tumor cell injection, every 3 days for 2 weeks. Notably, none of the mice succumbed during the experimental process. At the end of the experiment, the mice were euthanized using CO₂ (100% CO₂, 20 l/min; total chamber volume, 40 l; displacement rate, 50% l/min), and to verify death, breathing was assessed. In the present study, all nude mice showed a maximum weight loss of <15% before euthanasia.

After tumor formation, animals were euthanized, and tumor tissues were harvested. The tumor tissues were fixed in 4% paraformaldehyde at 4°C for 24 h and 4-μm paraffin-embedded sections were prepared. The paraffin-embedded sections were dehydrated in a descending series of ethanol, followed by antigen repair with sodium citrate at 95-100°C, and blocking with H₂O₂ at room temperature for 5-10 min and goat serum (cat. no. 16210064; Thermo Fisher Scientific, Inc.) at room temperature for 10 min. The sections were then stained with an anti Ki-67 antibody (1:100; cat. no. TA500265; OriGene Technologies, Inc.) at 4°C overnight and with a biotin-labeled secondary antibody (cat. no. SAP-9100; OriGene Technologies, Inc.) at 37°C for 30 min. After washing, Ki-67 staining was observed under an optical microscope, and image analysis software (ImageJ v1.8.0; National Institutes of Health) was used to semiquantitatively evaluate the intensity and degree of Ki-67 staining.

The aforementioned paraffin-embedded tumor tissue sections were subjected to apoptosis detection using the TUNEL cell apoptosis assay kit (cat. no. C1091; Beyotime Institute of Biotechnology) according to the manufacturer's instructions. Under an optical microscope, five fields of view were observed. And the brown yellow staining represented apoptotic cells. Apoptosis rate was calculated as number of positive cells/total number of cells.

GraphPad Prism 5 software (Dotmatics) was used for statistical analysis. Each experiment was performed at least in triplicate, and the data are presented as the mean ± SD. Comparisons between two groups were performed using a two-tailed unpaired Student's t-test, and multiple groups were analyzed by one-way ANOVA followed by Tukey's post hoc test or mixed ANOVA followed by Bonferroni's post hoc test. P<0.05 was considered to indicate a statistically significant difference.

RT-qPCR and western blotting were performed to measure the expression levels of YTHDF1 in glioma tissue and NBT. RT-qPCR showed that the mRNA expression levels of YTHDF1 were significantly upregulated in glioma tissue (Fig. 1A). Consistent with the trend in gene expression, the protein expression levels of YTHDF1 in glioma tissues were also significantly higher than those in NBTs (Fig. 1B and C). Subsequently, RT-qPCR was conducted to measure the expression levels of LINC00900 in glioma tissue and NBT. As shown in Fig. 1D, the expression levels of LINC00900 in glioma tissues were significantly higher than those in NBTs. A total of 16 patients had GBM (WHO grade IV) (26) and 4 patients had astrocytoma (WHO grade III). As shown in Fig. S2, LINC00900 expression in the glioma tissues of the two groups was significantly higher than that in the adjacent tissues, and no significant difference was found in the expression of LINC00900 between the two groups. Subsequently, MeRIP-PCR analysis revealed that the LINC00900 transcript in glioma tissues was enriched with m6A (Fig. 1E), suggesting that the high content of m6A in LINC00900 transcripts may be the cause of its abnormal expression in glioma tissue.

Four glioma cell lines (A172, U87, U251 and SHG44) were selected to induce GSCs. GSCs were generated by serum-free induction and, as shown in Fig. S3A, stem cells were successfully induced and named GSC-U87 and GSC-U251. GSC-A172 and GSC-SHG44 cells were also induced and are shown in Fig. S4A. The GSCs were identified by immunofluorescence (Figs. S3B and S4B), and western blotting (Fig. S3C) of the stem cell markers CD133, NANOG and OCT4. RT-qPCR and western blotting showed that compared with that in A172, U87, U251 and SHG44 cells, YTHDF1 expression levels in GSC-A172, GSC-U87, GSC-U251 and GSC-SHG44 cells were significantly increased (Fig. 2A and B). U251 and U87 cells are two commonly used glioma cell lines for in vitro experiments, which were isolated from GBM and are often used as an in vitro model of GBM. Therefore, these cell lines were selected to study the effects of YTHDF1 on the viability, invasion and self-renewal of GSCs. Subsequently, GSC-U87 and GSC-U251 cells were infected with YTHDF1 overexpression or knockdown lentiviruses. As shown in Fig. 2C-E, YTHDF1 overexpression significantly upregulated YTHDF1 mRNA and protein expression levels, whereas YTHDF1 knockdown significantly downregulated YTHDF1 mRNA and protein expression levels in GSC-U87 and GSC-U251 cells. Notably, shRNA-2 had the best knockdown efficiency, and was therefore used in subsequent experiments. The results of a CCK-8 assay showed that YTHDF1 overexpression significantly increased GSC-U87 and GSC-U251 cell viability, whereas knockdown of YTHDF1 significantly decreased GSC-U87 and GSC-U251 cell viability (Fig. 2F). The Transwell assay showed that YTHDF1 overexpression significantly increased the invasive ability of GSC-U87 and GSC-U251 cells, whereas knockdown of YTHDF1 significantly decreased the invasive ability of GSC-U87 and GSC-U251 cells (Fig. 2G and H). The tumor sphere formation assay showed that YTHDF1 overexpression significantly increased the tumor sphere formation ability of GSC-U87 and GSC-U251 cells, whereas knockdown of YTHDF1 significantly decreased the tumor sphere formation ability of GSC-U87 and GSC-U251 cells (Fig. 2I and J). These results suggested that YTHDF1 overexpression promoted the viability, invasion and self-renewal of GSCs.

To investigate whether YTHDF1 affects the expression of LINC00900, YTHDF1 overexpression or knockdown was performed in GSC cells. The results showed that the expression levels of LINC00900 were increased in YTHDF1-overexpressing GSCs and were decreased in YTHDF1-knockdown GSCs (Fig. 3A and B). RNA pull-down and RIP assays were used to demonstrate the binding of the YTHDF1 protein to LINC00900 in GSCs. As shown in Fig. 3C and D, the RNA pull-down assay showed that the biotin-labeled LINC00900 (bio-LINC00900) pull-down complex contained YTHDF1 protein. Moreover, in the RIP assay, LINC00900 was detected in the YTHDF1 antibody pull-down complex. In addition, there was a significant difference between anti-YTHDF1 and IgG groups, indicating that the enrichment of LINC00900 on the YTHDF1 protein was significantly increased (Fig. 3E and F). These results identified the binding relationship between LINC00900 and YTHDF1. Actinomycin D can be absorbed by cells within a few minutes and preferentially embedded into DNA sequences rich in GC, forming stable complexes that inhibit the transcription process of all eukaryotic RNA polymerase enzymes. After treating with actinomycin D for different durations, the molecular level of target RNA (such as mRNA and lncRNA) can be detected, the half-life of the target RNA can be calculated, and its stability evaluated. Using actinomycin D, it was revealed that LINC00900 expression was significantly reduced in YTHDF1-knockdown GSCs (Fig. 3G and H), indicating that YTHDF1 affected LINC00900 expression by regulating its stability.

To further explore whether YTHDF1 regulates LINC00900 expression and affects GSC functions, YTHDF1 was overexpressed and LINC00900 was knocked down. Fig. S5A and B shows successful knockdown of LINC00900 in GSC-U87 and GSC-U251 cells. The results of a CCK-8 assay showed that YTHDF1 overexpression promoted viability, whereas knockdown of LINC00900 reversed the enhancing effect of YTHDF1 on viability (Fig. 4A). The Transwell assay showed that knockdown of LINC00900 reversed the effect of YTHDF1 on GSC invasion (Fig. 4B and C). The tumor sphere formation assay showed that knockdown of LINC00900 reversed the effect of YTHDF1 on GSC tumor sphere formation (Fig. 4D and E). These data indicated that YTHDF1 promoted GSC viability, invasion and self-renewal by upregulating LINC00900 expression.

Prediction analysis using LncBase v.2 revealed a miR-1205-binding site on the LINC00900 transcript (Fig. 5A). miR-1205 expression was revealed to be significantly decreased in glioma tissue compared with that in NBT (Fig. S6A). The luciferase reporter assay confirmed the target binding relationship between miR-1205 and LINC00900 (Fig. 5B). Subsequently, LINC00900 was successfully overexpressed in GSC-U251 and GSC-U87 cells (Fig. S5A and B). The results revealed that LINC00900 overexpression significantly inhibited miR-1205 expression, whereas knockdown of LINC00900 upregulated miR-1205 expression (Fig. 5C and D), implying that LINC00900 negatively regulated miR-1205 expression.

Prediction analysis using TargetScan also revealed a miR-1205-binding site on the STAT3 3′-UTR (Fig. 5E). The mRNA and protein expression levels of STAT3 were significantly increased in glioma tissue compared with those in NBT (Fig. S6B and C). The luciferase reporter assay confirmed the target binding relationship between miR-1205 and the STAT3 3′-UTR (Fig. 5F). In addition, it was revealed that LINC00900 overexpression significantly upregulated STAT3 expression, whereas knockdown of LINC00900 inhibited STAT3 expression (Fig. 5G-I), implying that LINC00900 positively regulated STAT3 expression. Furthermore, LINC00900 and miR-1205 were overexpressed in GSCs. Fig. S5C and D show successful overexpression of miR-1205 in GSC-U87 and GSC-U251 cells. The miR-1205 mimics reversed the upregulating effect of LINC00900 on STAT3 expression (Fig. 5J), which indicated that LINC00900 may upregulate STAT3 expression by sponging miR-1205.

To further explore whether LINC00900 regulates the miR-1205/STAT3 axis and thus affects GSC functions, LINC00900 and miR-1205 were overexpressed in GSCs. The transfection efficiency of LINC00900 and miR-1205 is shown in Fig. S5. The results of a CCK-8 assay showed that LINC00900 overexpression promoted viability, whereas the miR-1205 mimics reversed the inducing effect of LINC00900 on cell viability (Fig. 6A). The Transwell assay showed that the miR-1205 mimics reversed the effect of LINC00900 on GSC invasion (Fig. 6B and C). The tumor sphere formation assay showed that the miR-1205 mimics also reversed the effects of LINC00900 on GSC tumor sphere formation (Fig. 6D and E).

Analysis of xenografted tumors in nude mice showed that inhibition of YTHDF1 or LINC00900 slowed tumor growth rate, and the tumor volume and weight of the simultaneous YTHDF1 and LINC00900 inhibition group were lower than those in the single inhibition groups (Fig. 7A-C). Compared with tumors in the LV-sh NC1 + LV-sh NC2 group mice, miR-1205 expression was increased in tumors in the LV-sh-YTHDF1 and LV-sh-LINC00900 group mice, whereas STAT3 expression was reduced (Fig. 7D and E). Additionally, compared with tumors in the LV-sh NC1 + LV-sh NC2 group mice, Ki-67 expression in tumors in the LV-sh-YTHDF1 and LV-sh-LINC00900 group mice was decreased (Fig. 7F and G), whereas the proportion of TUNEL-positive cells was increased (Fig. 7H and I). These findings indicated that LINC00900 promoted GSC viability, invasion, self-renewal and tumor growth by regulating the miR-1205/STAT3 axis.