Overexpression of FOXD2‐AS1 enhances proliferation and impairs differentiation of glioma stem cells by activating the NOTCH pathway via TAF‐1

Abstract Emerging data have highlighted the importance of long noncoding RNAs (lncRNAs) in exerting critical biological functions and roles in different forms of brain cancer, including gliomas. In this study, we sought to investigate the role of lncRNA FOXD2 adjacent opposite strand RNA 1 (FOXD2‐AS1) in glioma cells. First, we used sphere formation assay and flow cytometry to select U251 glioma stem cells (GSCs). Then, we quantified the expression of lncRNA FOXD2‐AS1, TATA‐box binding protein associated factor 1 (TAF‐1) and NOTCH1 in glioma tissues and GSCs, as well as the expression of GSC stem markers, OCT4, SOX2, Nanog, Nestin and CD133 in GSCs. Colony formation assay, sphere formation assay, and flow cytometry were used to evaluate GSC stemness. Next, the correlations among lncRNA FOXD2‐AS1, TAF‐1 and NOTCH1 were investigated. LncRNA FOXD2‐AS1, TAF‐1 and NOTCH1 were found to be elevated in glioma tissues and GSCs, and silencing lncRNA FOXD2‐AS1 inhibited stemness and proliferation, while promoting apoptosis and differentiation of GSCs. LncRNA FOXD2‐AS1 overexpression also led to increased NOTCH1 by recruiting TAF‐1 to the NOTCH1 promoter region, thereby promoting stemness and proliferation, while impairing cell apoptosis and differentiation. Mechanistically, lncRNA FOXD2‐AS1 elevation promoted glioma in vivo by activating the NOTCH signalling pathway via TAF‐1 upregulation. Taken together, the key findings of our investigation support the proposition that downregulation of lncRNA FOXD2‐AS1 presents a viable and novel molecular candidate for improving glioma treatment.


| INTRODUC TI ON
Gliomas are the most commonly occurring brain tumours in adults, which exhibit different degrees of malignancy and histological subtypes. 1 High-grade gliomas are the major cause of brain tumour death. 2 Similar to other cancers, recent evidence has demonstrated that glioma stem cells (GSCs) show tumourinitiating properties in malignant gliomas. 3 Radiotherapy is among the few treatments for patients with gliomas, but many gliomas become largely radio-resistant and almost inevitably recur after treatment. 4 Aside from radiotherapy, anti-angiogenesis drugs against tumour cells exhibits therapeutic effects against gliomas.
The anti-angiogenesis drug known as bevacizumab is an antibody that acts against vascular endothelial growth factor, resulting in decreased malignant glioma growth. 5 However, despite these approaches and their therapeutic effects, the treatment of malignant gliomas still remains palliative.
It is generally accepted that long noncoding RNAs (lncRNAs) can exert critical biological functions in cancers including brain tumours. 6 LncRNAs are transcripts containing more than 200 nucleotides that are not translated into their protein products. 7 Despite initial findings that lncRNAs play a minor physiological role despite their great abundance, an increasing amount of current research suggests an important participation in cancer biology. 8,9 Indeed, the progression and development of gliomas were found to be regulated by several lncRNAs. 10 For example, lncRNA nuclear enriched abundant transcript 1 has been indicated to promote glioblastoma cell growth and invasion by increasing β-catenin nuclear transport. 11 In addition, lncRNA paternally expressed 10 has been demonstrated as a biomarker for relapse and oncologic outcomes of patients with gliomas. 12 As revealed from the bioinformatics analysis that we conducted prior to the present investigation, lncRNA FOXD2 adjacent opposite strand RNA 1 (FOXD2-AS1) is a notably upregulated lncRNA in glioma samples. More importantly, the ectopic expression of lncRNA FOXD2-AS1 is indicative of poor prognosis for patients with nasopharyngeal carcinoma. 13 The contribution of lncRNA FOXD2-AS1 to the progression of glioma has also been documented previously. 14,15 Interestingly, the NOTCH signalling pathway was reported to be regulated by lncRNA FOXD2-AS1 in colorectal cancer. 13 The NOTCH signalling pathway is classically known for its role in determining cell fate during the development of the embryo and mature tissues. 16 In addition, NOTCH1 signalling pathway activation can maintain the stem cell phenotype of glioma initiating cells. 17 The inactivation of NOTCH signalling pathway was proved to be an effective target for glioma stem therapy. 18 Furthermore, in our preliminary research in silico, we found a correlation between lncRNA FOXD2-AS1 and TATA-box binding protein associated factor 1 (TAF-1), based on the LncMAP database.
TAF-1 protein is a pivotal component of the transcription factor II D complex, which plays a key role on transcription initiation. 19 As previously reported, LINC00319 directly binds to TAF-1 and further regulates HMGA2 in gliomas, 20 leading us to hypothesize that lncRNA FOXD2-AS1 is involved in glioma via TAF-1 and the NOTCH signalling pathway. In our study, we aimed to explore the specific role of lncRNA FOXD2-AS1 in the proliferation and differentiation of GSCs along with the relationship between lncRNA FOXD2-AS1/TAF-1/NOTCH1 signalling pathway in glioma. and of lncRNA FOXD2-AS1 with NOTCH1 were analysed, and scatter plots were drawn.

| Clinical sample collection
Glioma tissues were collected from 26 patients (14 males and 12 females; average age of 45.65 years, ranging from 20 to 71 years) following surgical operations in Taihe Hospital from January 2013 to May 2016. Glioma tissues were stored at −80 °C after collection.
A total of 26 patients who underwent intracranial decompression in our hospital during the same period were selected as the control group with their normal brain tissues collected. After surgery, the patients received concurrent radiotherapy of temozolomide combined with adjuvant chemotherapy, which is the standard regimen for such patients (i.e. Stupp regimen). The detailed demographic information of all patients is shown in Table S1. To isolate GSCs, U251 cells were cultured in a serum-free medium (Hyclone, Logan, UT, USA) containing epidermal growth factor, basic fibroblast growth factor, leukaemia inhibitory factor and melanocortin 2 receptor accessory protein (final concentration of 20 ng/mL; Peprotech EC, London, UK). The suspended tumour cell spheres were then separated by immunomagnetic beads to obtain CD133 positive cells (CD133 + cells magnetic bead sorting Kit;
HEK293T cells were transfected with different plasmids, and supernatants were collected after a 48-hour culture period. Virus in the supernatants were then filtered and centrifuged, until the desired viral titre was obtained. GSCs in the logarithmic growth phase were collected and prepared into single cell suspensions containing 5 × 10 4 cells/mL, followed by incubation in 6-well plates with 2 mL in each well at 37 °C overnight. GSCs were then infected with lentivirus expressing short hairpin (sh)-NC, overexpression (oe)-NC, sh-lncRNA FOXD2-AS1, oe-lncRNA FOXD2-AS1, oe-TAF-1, sh-TAF-1 or oe-FOXD2-AS1 + sh-TAF-1. After 48 hours, reverse transcription quantitative polymerase chain reaction (RT-qPCR) was used to determine expression of related genes.

| RNA extraction and quantification
An RNA extraction kit (Qiagen, Hilden, Germany) was used to extract the total RNA from glioma tissues and GSCs. ReverTra Ace ® qPCR RT Master Mix with gDNA Remover Kit (Toyobo, Ohtsu, Shiga, Japan) was used to reversely-transcribe RNA into cDNA according to the manufacturer's instructions. qPCR was then performed using Bio-Rad CFX96 Touch™ fluorescence detection system according to instructions provided in a 2 × RealStar Green Mixture kit (GenStar BioSolutions, Beijing, China). Glyceraldehyde phosphate dehydrogenase (GAPDH) was used as internal reference. The fold change was determined using the relative quantification (2 −ΔΔCt method) as previously described. 22 The primers were synthesized in Shanghai Sangon Co., Ltd. (Shanghai, China) (Table S2).

| Sphere formation assay
Single cell suspension was seeded onto ultra-low attachment sixwell plates (Corning Inc., Corning, NY, USA) at a density of 5 × 10 3 cells/mL and cultured in modified serum-free DMEM. The medium was replaced every three days. After 14 days, spheres at passage 1 with a size >100 µm were counted. Then, for the secondary sphere formation, the spheres were scattered and re-seeded in 96-well plates at a density of approximately 100 cells per well with the number of secondary spheres calculated 10 days after incubation.

| Dual-luciferase reporter gene assay
Luciferase reporter plasmids were co-transfected with oe-NC, oe-lncRNA FOXD2-AS1, sh-NC and sh-lncRNA FOXD2-AS1, respectively, to determine whether lncRNA FOXD2-AS1 could activate the NOTCH1 promoter. Renilla luciferase was used as internal normalization. After a transfection period of 48 hours, cells were collected and lysed. Luciferase Assay Kit (K801-200, BioVision, Hannover, Germany) was used to perform dualluciferase reporter assay in a dual fluorescent reporter analysis system (Promega, Madison, WI, USA). The activation of targeting reporter genes was determined by the ratio of relative light unit (RLU) of firefly luciferase to RLU of renilla luciferase. NOTCH1 promoter was constructed into a luciferase reporter vector cotransfected along with TAF-1 overexpressing plasmid into glioma cells to verify the binding relationship between TAF-1 and NOTCH1 promoter.

| RNA pull-down assay
Biotin-labelled lncRNA FOXD2-AS1 RNA and its anti-sense sequence were synthesized using Pierce RNA 3'desthiobiotinylation Kit (Thermo Fisher Scientific). After cells were lysed, the protein concentration (not lower than 2 mg/mL) was determined as above.
A magnetic RNA-Protein Pull-Down kit (Thermo Fisher Scientific) was used for the subsequent procedure. A bead suspension (50 μL) was added into an Eppendorf tube, and the magnetic beads were adsorbed on the magnetic bead sorter, washed with 20 mM Tris (pH = 7.5) and then re-suspended. After repeating the aforementioned procedure, the magnetic beads were adsorbed on the magnetic bead sorter again and re-suspended with 100 μL of 1 × RNA Capture Buffer. The magnetic sorter was used to absorb magnetic beads, and then, the beads were washed with 20 mM Tris (pH = 7.5), followed by re-suspension. The above procedures were repeated. A total of 100 μL 1 × Protein-RNA-Binding Buffer were added to the beads, followed by re-suspension and elution. The cell lysate was incubated with 1 pmol biotin-labelled lncRNA FOXD2-AS1 RNA and magnetic beads for 30-60 min at 4°C. Magnetic beads were absorbed and then washed with 100 μL RNA immunoprecipitation (RIP) Wash Buffer three times, followed by the addition of 50 μL Elution Buffer and incubated for 30 to 60 minutes at 4°C. LncRNA FOXD2-AS1 RNA-binding proteins were eluted next. After determining the protein concentration, Western blot analysis was used to measure the relative protein levels.

| RIP
The RIP was performed using an EZ-Magna RIP RNA-Binding Protein Immunoprecipitation kit (Millipore, Germany) according to the manufacturer's instructions to analyse the binding capacity of lncRNA FOXD2-AS1 to TAF-1. GSCs were washed with ice-cold PBS twice and lysed with RIPA lysis buffer, followed by 10-min centrifugation at 14,000 rpm at 4°C. The supernatant was extracted and co-precipitated with antibodies. Beads (50 μL) was washed, resuspended in 100 μL RIP Wash Buffer and incubated with rabbit antibody to TAF-1 (Abcam). The magnetic beads-antibody complex was then re-suspended in 900 μL RIP Wash Buffer and incubated with 100 μL cell extract at 4 °C overnight. Beads-protein complex was collected on magnetic pedestals, and detached with protease K. The RNA was then extracted and RT-qPCR was adopted to detect the co-precipitated RNA.

| Chromatin immunoprecipitation assay
GSCs were initially cultured in a 10 cm diameter dish. When the cell number reached 1 × 10 6 , the medium was discarded. The cells were crosslinked with 1% formaldehyde for 10-min at 37°C. Crosslinking was stopped by leaving the GSCs on ice for 5 minutes. The GSCs were collected and re-suspended in 200 μL sodium dodecylsulphate lysis buffer, and chromatin fragmentation was performed by using ultrasound on ice. The supernatant was collected after 10-minute cell centrifugation at 14,000 ×g at 4°C followed by dilution with ChIP buffer containing protease inhibitors. Part of the supernatant was taken as input, and the rest of the supernatant was incubated with rabbit antibody to TAF-1 (Abcam) at 4°C overnight.
Cross-linked agar was used to precipitate the endogenous DNAprotein complex. The supernatant was discarded after centrifugation (1000 rpm; 1 min) at 4 °C, and elution buffer was then used to elute the DNA-protein complex. The eluted supernatant and input DNA were de-crosslinked in a water bath for 6 hours (65°C) after addition of 5 mol/L NaCl (20 μL). The DNA fragments were then collected following detachment using protease K. RT-qPCR was used to determine the relative levels of NOTCH1 promoter. Primers for NOTCH1 promoter are: Forward: CAGGGGAGACCCCCTATCC; Reverse: TGAAAGTTTTCAGAGGCCAAAAG.
The expression of CD133 was analysed using a BD FACSCalibur flow cytometer.

| Colony formation in soft agar
GSCs were collected and re-suspended at a density of 1 × 10 6 cells/L. Distilled water was used to prepare agarose of low solubility with a concentration between 0.7 and 1.2%. Then, 1.2% agarose was mixed with 2 × medium containing 2 × antibiotics and 20% calf serum. The 3 ml mixture was then added into an agar plate, and the bottom layer of agar was placed into an incubator. Next, 0.7% agarose was mixed with 2 × cell medium, followed by the addition of 0.2 mL cell suspension, and the mixture was then added into the agar plate. After the upper layer of agar had solidified, the plate was placed in a 37°C incubator with 5% CO 2 for 10-14 days. Colonies containing more than 50 cells were then counted under a light microscope.

| Intracranial glioma models
Serum-free Roswell Park Memorial Institute 1640 medium (Gibco) was used to re-suspend transfected GSCs to prepare 2 × 10 6 cells/

| Statistical analysis
All data were processed and analysed using SPSS 21.0 statistical software (IBM Corp., Armonk, NY, USA). Measurement data were expressed as mean ± standard deviation. If the data conformed to normal distribution and homogeneity of variance, a paired t-test was used to analyse tumour tissues and normal tissues. Unpaired t-test was adopted to analyse differences of other two groups. Differences between multiple groups were analysed by one-way analysis of variance (ANOVA), followed by a Tukey multiple comparisons post-test.
Comparisons among multiple groups at different time points were performed with two-way ANOVA, and tumour volume at different time points was analysed using repeated measures ANOVA, followed by Bonferroni's post hoc test. Kaplan-Meier was used to analyse survival, while Log-Rank was utilized to analyse the levels of significance in different groups. A value of p < 0.05 was considered statistically significant.

| LncRNA FOXD2-AS1 expression is elevated in glioma tissues and GSCs
After analysis of the GSE104267 microarray dataset by R language, 101 differentially expressed lncRNAs were obtained.
Among these, 19 were upregulated and 82 were downregulated.
Among the upregulated lncRNAs, lncRNA FOXD2-AS1 ( Figure 1A,   Fig. S1) was found previously to be associated with glioma progression. 14,23,24 According to GEPIA, lncRNA FOXD2-AS1 was highly expressed in gliomas from the TCGA database ( Figure 1B), and predicted poor survival rate of patients ( Figure 1C). Therefore, lncRNA FOXD2-AS1 was selected for the following study. To better understand the function and roles of lncRNA FOXD2-AS1 in gliomas, we first used RT-qPCR ( Figure 1D) to determine lncRNA FOXD2-AS1 expression in glioma tissues and normal tissues. We found that lncRNA FOXD2-AS1 was highly expressed in glioma tissues compared with normal tissues. Kaplan-Meier curves ( Figure 1E Figure 1F). Thus, U251 was selected for use in our study.
Taken together, we find that lncRNA FOXD2-AS1 expression was elevated in glioma tissues and cells.
The GSCs were then isolated from U251 cells, and a sphere formation assay was performed. The morphological characteristics of spherical U251 cell colonies are shown in Fig. S2A. Non-adherent cells were collected every three days, and RT-qPCR was used to determine expression of GSCs markers (OCT4, SOX2, Nanog, Nestin and CD133). As shown in Figure 1G, the expression of these genes gradually increased with time in culture. Immunofluorescence revealed that Nestin was positively expressed in GSCs, while almost no GFAP expression ( Figure 1H). Flow cytometry was used to determine the expression of CD133 in GSCs ( Figure 1I), which helped to confirm the successful isolation of GSCs. RT-qPCR was used to determine lncRNA FOXD2-AS1 expression in GSCs, which showed a high lncRNA FOXD2-AS1 expression in GSCs ( Figure 1J). These results indicated that lncRNA FOXD2-AS1 was upregulated in GSCs.

| Silencing lncRNA FOXD2-AS1 reduces stemness and proliferation but induces apoptosis and differentiation of U251 GSCs
To better elucidate the role of lncRNA FOXD2-AS1 in GSCs, lncRNA FOXD2-AS1 was silenced in U251 cells, and the stemness and differentiation of U251 GSCs were determined. The silencing efficiency of U251 cells treated with lncRNA FOXD2-AS1 shR- FOXD2-AS1#1 and sh-lncRNA FOXD2-AS1#2) was determined through RT-qPCR (Figure 2A). Results showed F I G U R E 1 LncRNA FOXD2-AS1 was highly expressed in glioma tissues and cells as well as GSCs. A, LncRNA FOXD2-AS1 expression predicted in GSE10 4267 (p = 1.482E-07). The box on the left represents expression of lncRNA FOXD2-AS1 in normal tissues and the box on the right represents expression in cancer tissues. B, A box plot displaying lncRNA FOXD2-AS1 expression in glioma samples (left) and normal samples (right) from TCGA database using GEPIA. C, Survival curve of patients with glioma regarding lncRNA FOXD2-AS1 expression from TCGA database using GEPIA (HR = 1.8, p = 0.032). D, LncRNA FOXD2-AS1 expression in glioma tissues and normal tissues detected by RT-qPCR (n = 26). E, A Kaplan-Meier one-way survival analysis according to lncRNA FOXD2-AS1 expression, followed by logrank test (n = 26). F, FOXD2-AS1 expression in glioma cell lines determined using RT-qPCR. G, Expression of lncRNA FOXD2-AS1 and GSC markers (OCT4, SOX2, Nanog, Nestin and CD133) determined using RT-qPCR. H, Expression of Nestin and GFAP in GSCs determined by immunofluorescence. I, CD133 + GSCs selected using flow cytometry. J, LncRNA FOXD2-AS1 expression in GSCs determined using RT-qPCR. * p < 0.05. Measurement data were expressed as mean ± standard deviation. Paired t-test was used to analyse comparison within group. Unpaired t-test was adopted to analyse differences between two groups. Differences among multiple groups were analysed by one-way ANOVA, followed by a Tukey multiple comparisons post-test. Comparisons among multiple groups at different time points were performed with two-way ANOVA. Cell experiments were conducted three times independently.  (Fig. S2B). Therefore, sh-FOXD2-AS1#1 was selected for subsequent functional assays. We then calculated the number of main spheres for every 1000 GSCs and the secondary sphere for every 100 GSCs, which was reduced significantly after treatment by sh-lncRNA FOXD2-AS1 ( Figure 2B).

Colony formation assay and flow cytometry helped to illustrate the impaired proliferation and enhanced apoptosis of U251
GSCs in the presence of sh-lncRNA FOXD2-AS1 ( Figure 2E,F).
Immunofluorescence staining was employed to determine the expression of astrocyte marker GFAP and stem cell marker CD133 in U251 GSCs. As depicted in Figure 2G,H, GFAP expression was significantly upregulated, while CD133 expression was downregulated in response to sh-lncRNA FOXD2-AS1. These findings suggested that silencing lncRNA FOXD2-AS1 reduced stemness and proliferation, but promoted apoptosis and differentiation of U251 GSCs.

| LncRNA FOXD2-AS1 elevates NOTCH1 by recruiting TAF-1 to promoter region of NOTCH1 in GSCs
The interaction of lncRNA FOXD2-AS1 and TAF-1 with NOTCH1 was found based on the LncMAP database, and TAF-1 was initially predicted to be highly expressed in glioma according to the GEPIA website ( Figure 3A). Moreover, the activation of NOTCH signalling pathway was previously found to be associated with GSC differentiation and metastasis. 17,25,26 We therefore hypothesized that lncRNA FOXD2-AS1 could potentially activate the NOTCH signalling pathway by regulating NOTCH1 via TAF-1. To test this hypothesis, RT-qPCR was initially adopted to determine TAF-1 expression in glioma tissues. TAF-1 expression was found to be higher in glioma tissues than in normal tissues ( Figure 3B). Moreover, RT-qPCR and Western blot analysis revealed that the level of TAF-1 was higher in U251 GSCs ( Figure 3C and Fig. S3A). As shown in Figure 3D,E, lncRNA FOXD2-AS1 bound to TAF-1 in U251 GSCs, which was further verified using RIP and RNA pull-down assays.
We initially predicated that multiple binding sites existed between TAF-1 and NOTCH1 ( Figure 3F) according to the GTRD website.
A ChIP assay revealed that TAF-1 could indeed bind to NOTCH1 at its promoter region ( Figure 3G). RT-qPCR and Western blot analysis depicted that NOTCH1 expression in glioma tissues was much higher than that in normal tissues ( Figure 3H and Fig. S3B).
Pearson's correlation analysis also found a positive relationship between the expression of lncRNA FOXD2-AS1 and NOTCH1, as well as that between the expression of TAF-1 and NOTCH1 ( Figure 3I).
Dual-luciferase reporter gene assays were carried out to help verify the relationship between lncRNA FOXD2-AS1 and NOTCH1 promoter. The results revealed that the NOTCH1 promoter was activated in response to oe-lncRNA FOXD2-AS1 treatment, while the opposite occurred in response to sh-lncRNA FOXD2-AS1. This strongly suggested that lncRNA FOXD2-AS1 could specifically activate NOTCH1 ( Figure 3J). ChIP assay ( Figure 3K

| Overexpression of lncRNA FOXD2-AS1
activates the NOTCH signalling pathway by recruiting

TAF-1, thereby promoting GSC stemness and repression of GSC differentiation
Since lncRNA FOXD2-AS1 was able to upregulate NOTCH1, we decided to explore whether the entire NOTCH pathway could be similarly activated. We first conducted a Western blot analysis to determine expression of NOTCH pathway related genes (JAG1, PS1 and HES1). As described in Figure 4A, JAG1, PS1 and HES1 expression in GSCs were significantly elevated in response to oe-lncRNA FOXD2-AS1, which was reversed by silencing TAF-1.
The number of main spheres in every 1000 GSCs and the secondary spheres of every 100 GSCs was increased in response to oe-lncRNA FOXD2-AS1, which was restored following treatment of oe-lncRNA FOXD2-AS1 + sh-TAF-1 ( Figure 4B). OCT4, SOX2, Nanog, Nestin and CD133 expression in GSCs assessed by RT-qPCR demonstrated that the expression of these markers was significantly enhanced after overexpression of lncRNA FOXD2-AS1, but that these effects were abrogated by treatment with oe-lncRNA FOXD2-AS1 + sh-TAF-1 ( Figure 4C). Flow cytometry ( Figure 4D) revealed an obvious increase in the number of CD133 + cells in U251 GSCs in response to oe-lncRNA FOXD2-AS1 treatment while this effect was rescued after transfection with oe-lncRNA FOXD2-AS1 + sh-TAF-1.

| DISCUSS ION
Gliomas are categorized into five principal groups based on tumour markers and into different subgroups according to their underlying pathogenesis. 30 The manifest diversity of different types of gliomas poses a pressing challenge in their treatment. Malignant gliomas constitute as the most aggressive subtype of gliomas. There remains a lack of effective therapeutic treatment regimen, which is due in part to the unhampering proliferation of cancer stem cells, which is a known aetiology of malignant gliomas. 31 Currently, the specific molecular mechanism underlying GSS proliferation and dif-   In our study, we also found that lncRNA FOXD2-AS1 overexpression was able to induce GSC stemness and proliferation as well as inhibiting GSC differentiation and apoptosis through activation of the NOTCH signalling pathway by recruiting TAF-1, as evidenced by increased expression of OCT4, SOX2, Nanog, Nestin and CD133.
Previous work showed that the expression of OCT4, SOX2 and Nanog correlated positively with tumour malignancy in human gliomas. 35 In addition, a previous study also noted that CD133 and Nestin were cancer cell stemness markers, whereby CD133 and Nestin overexpression was correlated with worse prognosis in terms of overall survival. 36 A recent study found that lncRNA FOXD2-AS1 upregulation resulted in larger tumour size, greater invasion depth, distant metastasis and advanced tumour node metastasis staging in cancer. 37 Moreover, the knockdown of lncRNA FOXD2-AS1 significantly suppressed proliferation, and promoted apoptosis of glioma cells. 23 The repression of lncRNA FOXD2-AS1 was reported to reduce proliferation, migration, invasion, stemness of glioma cells and impaired tumour growth in transplanted tumours, 24 which was consistent to what was found in our study. TAF-1 was previously found to be functional in glioma targeted by lncRNA LINC00319. 20 We found that NOTCH1 was expressed at higher levels in GSCs and glioma tissues, consistent with results in a prior study. 38 The NOTCH signalling pathway plays a pivotal role on the self-renewal, proliferation and differentiation of neural stem cells or neural precursor cells. 39 In addition, NOTCH1 downregulation exhibits inhibitory effects on GSC proliferation, 29 which has been regarded to present an effective strategy for GSC-targeting therapy. 18 Therefore, in our study, we demonstrated that lncRNA FOXD2-AS1 could activate the NOTCH signalling pathway. Likewise, lncRNA FOXD2-AS1 knockdown inactivated NOTCH signalling pathway to suppress colorectal cancer development. 40

| CON CLUS ION
This study provides new insights into the molecular mechanism of how lncRNA FOXD2-AS1 is involved in the pathogenesis of gliomas.
LncRNA FOXD2-AS1 is upregulated in GSCs and glioma tissues, which helps to positively regulate NOTCH1 expression. Increased levels of the lncRNA FOXD2-AS1 also upregulate NOTCH1 by recruiting TAF-1 to the NOTCH1 promoter region, thereby promoting GSC stemness and proliferation as well as repressing GSC differentiation and apoptosis ( Figure 6). The mechanism elucidated in our investigation may present new therapeutic targets for treating gliomas.

CO N S E NT FO R PU B LI C ATI O N
Not applicable.

This study was supported by Scientific Research Project of Health
Commission of Hubei Province (No. WJ2019F059).

CO N FLI C T S O F I NTE R E S T
The authors declared that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.