NSG-70, a new glioblastoma cell line with mixed proneural-mesenchymal features, associates NOTCH1-WNT5A signaling with stem cell maintenance and angiogenesis

Glioblastoma initiation and progression is believed to be driven by Glioma stem cells (GSCs). Activation of NOTCH1 and WNT, and more recently, non-canonical WNT5A signaling, has been demonstrated to regulate self-renewal and differentiation of the GSCs crucially. High expression levels of NOTCH1 and WNT in GBM tumors contribute to the sustenance of GSCs and mediate characteristic phenotypic plasticity, which is reflected by the different subtypes and tremendous intra-tumor heterogeneity. However, the coregulation of NOTCH1 and WNT5A is not well understood. Here, we studied the role of these molecules in regulating the characteristics of different GSC subtypes. We established a novel GSC-enriched cell model, referred to as NSG-70, from a patient with recurrent GBM. NSG-70 cells harbor a unique cytogenetic feature, viz. isochromosome 9q. At the same time, its expression profiles indicate that it is a mixed lineage comprising proneural and mesenchymal subtypes. We examined the relevance of NOTCH1 and WNT5A signaling and their coordinated action in GBM using these cells and other patient-derived models representing different GSC subtypes. Our data revealed that the downregulation of NOTCH1 resulted in the suppression of stem cell and mesenchymal markers and significantly reduced the levels of WNT5A. NOTCH1 knockdown also led to a notable reduction in the vasculogenic mimicry of GSCs. Interestingly, knockdown of WNT5A exhibited similar effects and drove quiescent GSC towards proliferation. In a complementary manner, ectopic expression of WNT5A or rhWNT5A treatment rescued the effects of NOTCH1 knockdown. The resistance of GSCs towards conventional therapies in part due to subtype interconversion demands therapies targeting specific GSC subtype. Our study suggests the need for a combinatorial approach that could effectively target the NOTCH1-WNT5A signaling axis toward eliminating GSCs.


Introduction
Glioblastoma (GBM) is the most common and malignant tumor of the central nervous system (CNS). It exhibits extensive cellular and molecular heterogeneity, which is responsible for treatment failure and tumor recurrence [1]. Several pieces of evidence suggest that GBM is driven by a subpopulation of cancer cells with stem cell characteristics and which possess a potential for self-renewal and multipotentiality [2,3]. The survival and self-renewal properties of glioma stem cells (GSCs) are stringently dependent on communication between stem cells and their surrounding cells (stem cell niche), including endothelial and differentiated tumor cells, via several tightly regulated signaling pathways [4]. A major driving force for regulating self-renewal and proliferative properties of GSCs is mediated by Notch and Wnt signaling [5].
GBM tumors are categorized into four molecular subgroups, viz. classical, neural, proneural, and mesenchymal, based on dominant gene expression patterns in each group and are assigned putative prognostic significance [6]. However, most studies identify and characterize two mutually exclusive GSC subtypes: proneural and mesenchymal, with the mesenchymal subtype being more aggressive than proneural [7,8]. GSCs of all subtypes exhibit phenotypic plasticity, further contributing to the intratumoral heterogeneity [9], besides exhibiting dysregulation of distinct signaling pathways [8], which needs to be studied in detail. Hence, a better understanding of GSC subtypes and the signaling pathways maintaining their stemness is indispensable in developing targeted and personalized therapies.
The Notch and canonical Wnt signaling pathways are shown to majorly modulate the motility and invasiveness of hGBM cells by activating an epithelial to mesenchymallike program in malignant gliomas [10]. They also play an indispensable role in the proliferation of GSCs and neoangiogenesis [11]. The notch signaling pathway is implicated in the maintenance and proliferation of the neural stem cells (NSCs) as well as GSCs [12]. However, how the Notch pathway fine-tunes its functional mechanisms to bring about the differentiation process or maintenance of stemness in NSCs or GSCs, is largely unknown. Previous reports suggested that several pathways can crosstalk and modulate Notch signaling to suppress its effects on the self-renewal of NSCs/ GSCs and induce the differentiation of normal cells or tumor cells [12,13]. In association with Notch, Wnt/βcatenin signaling regulates core processes in stem and progenitor cells that include self-renewal, proliferation, differentiation, and migration [14,15]. Dysregulated Wnt/βcatenin signaling is frequently observed in several cancers, including GBM [15]. Previous reports showed the role of Wnt5a in promoting vasculogenic mimicry and angiogenesis in other cancers [16,17]. Recently, WNT5A has been shown to drive GSC differentiation into endothelial-like cells, which correlates with an increase in tumor neovascularization in GBM [18]. However, to our knowledge, there are no reports of crosstalk or coordinated action of NOTCH1 and non-canonical WNT ligand, WNT5A in GSCs. There is evidence indicating overexpression of WNT5A in high-grade gliomas, and a recent study also showed that Wnt5a overexpression induces an infiltrative, mesenchymal-like characteristic in glioma stem-like cells [19]. Moreover, unpublished data in the lab wherein screening of glioma cell-derived extracellular vesicles showed the presence of NOTCH1 and WNT5A led us to escalate the studies supporting the involvement of these pathways in regulating GSC stem cell phenotype and vasculogenic mimicry.
Here, we established and characterized a novel GSC line, referred to as NSG-70, which represents an in vitro model of mixed (proneural and mesenchymal) subtype. Karyotype analysis of NSG-70 GSC displayed several commonly reported chromosomal aberrations along with a unique feature of isochromosome 9q (i9q). Microarray analysis of NSG-70 confirmed the presence of i9q and gain of specific cytoband accommodating NOTCH1 gene, besides suggesting enhanced expression of additional genes, which may confer an advantage to tumor cell growth. This led us to study the biological effects of NOTCH1 signaling in NSG-70 along with two other pre-established GSC lines belonging to different subtypes from the lab to comprehensively identify the coregulation of NOTCH1 and WNT5A in modulating the stemness and proliferation potential of GSCs as well as in promoting angiogenesis. Targeting the Notch1-Wnt5A signaling axis could hence serve as an unprecedented treatment option for GBM.

Tumor tissue collection and cell culture
Tumor tissue and normal brain tissue specimens (from epileptic patients) were obtained following approval from the Institutional Ethics Committees (IEC) of NCCS, Pune, and KEM Hospital, Mumbai, India. Consent of patients was taken before the collection of tumor specimens. Tumor grading was performed by neuropathologists according to the World Health Organization classification. GSC cultures were established from tumor tissues of histopathologically verified glioblastoma patients. Tissues were mechanically and enzymatically dissociated into single cells and further plated in a serum-free medium supplemented with B27, EGF, and bFGF (20 ng/ml) for enriching stem cells as described earlier [20]. Culture flasks were coated with LDEV-free hESC qualified geltrex (Gibco) for 3 h at 10 ug/ 1 3 ml prior to use. Adherent GSC cultures were maintained in a serum-free medium as described earlier [21] and dissociated using Stempro Accutase (Gibco). The culture medium was changed twice a week, and cultures were observed under the microscope every 24-48 h. Primary cultures were propagated continuously up to 10-15 generations without any observed changes in cell morphology and growth characteristics. Human neural stem cell (hNSC) culture was derived from human induced pluripotent stem cells using Gibco pluripotent stem cell (PSC) Neural Induction Medium as per the manufacturer's instructions. The medium consists of Neurobasal® Medium and GIBCO® Neural Induction Supplement, 50X (Cat. no. A1647801).

Neurosphere formation assay
Adherent NSG-70 GSC culture was dissociated, and 500 cells were plated in 35 mm dishes. They were grown in the neurosphere culture conditions as described above for 12-14 days and were fed every 3-4 days by changing half of the medium. Once neurospheres reached approximately 200-300 cells, they were counted and presented as the percentage neurosphere forming cells of the total 500 cells plated. For limiting dilution assay, dissociated NSG-70 cells were seeded in a dilution range of 1-100 cells/well in a 96-well culture dish with a non-adherent bottom. They were allowed to grow and form floating spheroids consisting of 200-300 cells for 12-14 days with constant feeding of fresh media every 3 days. Spheroids were counted to obtain the percentage of neurosphere forming cells of the total cells plated.

Flow cytometry
Cell surface expression of CD133 and CD44 was examined by flow cytometry using a standard protocol. In brief, the adherent monolayer was dissociated by accutase treatment and harvested as a single-cell suspension. 2 μg/ml of phycoerythrin (PE)-conjugated anti-CD133, FITC-conjugated CD44 (Miltenyi, Auburn, CA), and negative control was added to 10 6 cells in 200 μl of immunofluorescence buffer (PBS containing 2% FBS and 0.02% sodium azide). After 1 h of incubation in the dark at 4 °C, 10,000 cells were analyzed using a Becton and Dickinson FACSCanto (Mountain View, CA). The results were processed using Diva software as the percentage of positive cells.

Tumorigenicity assay
Animal studies were approved by the Institutional Animal Ethics Committee (IAEC) of NCCS. Subcutaneous injections were performed in NOD-SCID mice with 10 4 , 10 5 , and 10 6 cells injected per mouse (3 mice per set), whereas stereotactic orthotopic injections were performed with 10 4 cells injected per mouse. At the end of the experiment, all mice were euthanized, and tumor tissues were collected. Finally, tumor volumes were calculated using the formula: a × b 2 /2, wherein 'a' and 'b' are the longest and the shortest diameters of the subcutaneous tumor, determined using a vernier caliper.

Pathology and immunohistochemistry
Hematoxylin and eosin (H&E) staining and immunohistochemistry were performed on NSG-70 and KW formalinfixed, paraffin-embedded (FFPE) tissue sections as described [23]. Antigen retrieval was performed by microwave treatment in citrate buffer (pH 6.0). Antibodies were used in prescribed dilution (Details in Supplementary Data), Vectastain Elite ABC kit (Vector Laboratories), and 3,3'-diaminobenzidine (Dako) was used as chromogen for the staining.

RNA extraction and quantitative real-time PCR
Total cellular RNA was extracted using TRIzol Reagent (Invitrogen) as per the manufacturer's instructions. Quantitative RT-PCR (qRT-PCR) was performed using Power-up SYBR Green (Applied Biosystems, Carlsbad, CA) using QuantStudio 6 Flex Real-Time PCR system (Thermo-Scientific). Primer sequences used are listed in Supplementary  Table 1.

Transcriptome analysis
Total RNA was isolated from the normal brain (NB) and three GSC cultures (KW, MUR, and NSG-70) using TRIzol (Invitrogen, Carlsbad, CA, USA). RNA (RIN > 7.0) was fragmented and then reverse transcribed by random hexamers and Superscript II Reverse Transcriptase (Life Technologies, Gaithersburg, Maryland, USA). Libraries from each sample were prepared by random fragmentation of cDNA followed by ligation, end repair, and addition. RNA concentrations and quality of each library were checked using a NanoDrop spectrophotometer (Thermo Scientific) and an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA), respectively. RNA sequencing of libraries was performed using Illumina HiSeq 2500 platform (Illumina, San Diego, CA) for paired-end sequencing. The paired-end reads were aligned to the reference human genome release downloaded from the UCSC database (GRCh37/hg19). The aligned reads were used for estimating the expression of the genes using program featureCounts (version: v1.5.2). The differential expression analysis was performed using DESeq2. Amigo2 GO annotation was performed for upregulated and downregulated genes. Transcripts with log2 fold change cutoff of 1 and p-value < 0.05 were considered as significantly deregulated genes.

Western blotting
Whole-cell lysates were prepared using RIPA protein extraction buffer with 1 × protease inhibitor cocktail (Sigma) following the manufacturer's instructions. Protein expression was detected using Immobilon Western Chemiluminescent HRP Substrate (Millipore). The details of the primary antibodies used are given in supplementary information.

Cell proliferation assay
Cell proliferation was determined by MTT assay as described previously. Briefly, 5 × 10 3 cells of each cell line per well were seeded in a 96 well plate. The absorbance was measured over a period of 5 or 6 days. 20 μL of MTT (5 mg/ mL) (Sigma-Aldrich) was added to each well, and readings were taken at 0 h and at every 24 h interval for 6 days. The formazan crystals formed were dissolved in dimethyl sulfoxide (DMSO) (Sigma-Aldrich), and the absorbance of each well was measured at OD-570 nm in a Bio-Rad microplate reader (Bio-Rad, Hercules, CA, USA).

Cell cycle analysis
1 × 10 6 GSCs were harvested and fixed by adding 70% ethanol drop-wise while vortexing and stored at 4 °C for at least 30 min. Cells were pelleted at 1200 rpm at room temperature for 5 min. Cells were rehydrated in PBS for 10 min after removing ethanol. GSCs were resuspended in 300 μL propidium iodide (PI)/Triton X-100 staining solution (PBS with 0.1% Triton X-100, 200 µg/mL DNase-free RNase A, 10 µg/ mL PI) and incubated for 30 min in dark. The samples were acquired in the BD FACSCalibur instrument with a 488 nm excitation laser. The analysis of cell cycle populations was performed in BD CellQuest Pro software.

Soft agar colony formation
The clonogenic potential of the GSC line was assessed by soft agar assay wherein 10 4 cells per well were seeded in 6-well plates in a 0.3% soft agar solution in containing complete GSC medium on top of a 1% agarose basal layer. The culture medium was also added to soft agar and refreshed every 3-4 days for 3 weeks. The cells were then fixed in 4% formaldehyde and stained with crystal violet for colony counting.

Generation of stable NOTCH1 and WNT5A knockdown cells
Three individual shRNAs to NOTCH1 and WNT5A were designed using the RNAi Consortium shRNA library, BROAD Institute, Cambridge, MA, and cloned into pLKO.1-TRC plasmid. lacZ sense and lacZ antisense sequence were used as the non-targeting control shRNA for the experiments. The shRNA plasmid constructs were confirmed by sequencing using human U6 promoter-specific forward primer and were designated as NOTCH1-shRNA1, 2, and 3; WNT5A-shRNA1, 2 and 3 and shlacZ (Control). The sequences of shRNAs designed for targeting NOTCH1 and WNT5A transcripts and non-targeting control shRNA are listed in Supplementary Table 2.
The HEK293T cell line was transfected using a 2nd generation lentiviral system for producing lentiviral vectors with Invitrogen Lipofectamine 3000 following the manufacturer's protocol. Lentiviral vectors (pMD2.G, Cat# 12,259 and psPAX2, Cat# 12,260) were purchased from Addgene Inc. (USA). Knockdown efficiencies of individual shRNA sequences were determined at RNA and protein levels (Supplementary Fig. S3c &S3d).
Out of 3 shRNAs used to knockdown NOTCH1, we obtained better knockdown efficiency with only shRNA2 in KW as well as NSG-70 GSC (

Generation of stable WNT5A overexpressing GSC lines
To generate stable cell lines, lentiviral transduced control GSCs (NSG-70-GFP and KW-GFP) and WNT5A overexpressing GSCs (NSG-70-WNT5A-GFP and KW-WNT5A-GFP) were sorted by flow cytometry (FACSAria). Three independent experiments were performed to obtain overexpressing GSC lines. Vectors (Sino Biologicals Inc.) used for lentiviral preparation are listed in the supplementary information.

Three-dimensional network formation on matrigel
Twenty-four-well tissue culture dishes were coated with 200 μL of Matrigel (BD Biosciences, USA) and incubated at 37 °C for 30 min. NSG-70 non-targeting shRNA, rhWNT5A treated NSG-70 non-targeting shRNA, NSG-70 NOTCH1 shRNA, and rhWNT5A treated NSG-70 NOTCH1 shRNA were suspended in a culture medium and seeded onto the plates at 6 X 10 4 cells per well. Cells were incubated for 24-36 h at 37 ℃, and VM networks were captured by phasecontrast microscopy followed by their quantitative analysis using AngioTool software. Three independent experiments were carried out for each condition, and 3 randomly selected fields were acquired per well [24].

Statistical analysis
All data were obtained from at least three replicates, and quantitative data are presented as mean ± standard deviation. Statistical analysis was performed using GraphPad Prism 5.0 software. Student's t-test was applied to compare differences between two groups, and one-way ANOVA followed by Dunnet's post hoc test was used for multiple comparisons. Statistical significance was denoted as: * p < 0.05, ** p < 0.01, *** p < 0.001 and **** p < 0.0001. All error bars represent the standard deviation of the mean. Unless otherwise mentioned, all experiments were performed at least in triplicate.

Characterization of NSG-70, a new glioblastoma stem cell line that harbours a marker isochromosome 9q (i9q)
NSG-70, a long-term cell line, was derived from tumor tissue of a patient with recurrent glioma and has been cultured through more than 25 passages, whereas concurrently established normal brain cell cultures (NB) derived from epilepsy patient-derived biopsies could not be maintained beyond 5-6 passages. The adherent NSG-70 culture presented as small, translucent, spindle-shaped cells characteristic of Grade IV gliosarcoma tumors [25,26] that could also generate nonadherent organized neurospheres within 8-10 days in the culture at a high frequency (55-60%), indicating a glioma stem cell (GSC)-like nature as represented by neurosphere with limiting dilution assay (Fig. 1a-ii and a-iii).
NSG-70 GSCs displayed a low proliferation potential with ~ 150 h doubling time (determined through MTT assay and low Ki-67 expression, Fig. S1a, b). This slowcycling (almost quiescent) nature was validated by PKH dye quenching assay that retained the dye in around 98.5% population even after six days in culture (Fig. S1c), and further reflected in high latency during xenograft formation, with all mice developing small tumors after four months of subcutaneous injection (10 6 cells; Fig. S1d) and one month of orthotopic injection (10 4 cells; Fig. S1f). Immunohistochemistry of the NSG-70 xenografts displayed typical spindleshaped morphology of the tumor cells that ascertained the gliosarcoma phenotype, along with wild type ATRX and IDH1, and mutated p53 expression ( Fig. 1b; Fig. S1e). Ki-67 and CD34 expression were restricted to 37.33% and 26.35% cells, affirming its low proliferation index and scanty tumor vessel density, respectively (Fig. 1b).
To ascertain their stem-like nature, we profiled NSG-70 cells for expression of neural progenitor (SOX2, NESTIN, and MSI1) as well as differentiation markers (glial fibrillary acidic protein -GFAP and S100 calcium-binding protein-S100β for astrocytic; TUJ1 and MAP2 for neuronal and OLIG2 for oligodendroglial lineages; Fig. 1c-i). Immunostaining of NSG-70 neurospheres also identified enhanced expression of stem cell markers CD133, SOX2, NESTIN, and MSI1, while flow cytometry-based profiling and quantification indicated the presence of ~ 18.6% CD133 and 99.8% CD44-positive cells in adherent cultures (Fig.,1d-i and 1d-ii). The presence of serum in the culture medium reduced the levels of stem cell marker expression (Fig. 1e). Xenograft tumors also expressed high levels of SOX2, CD44, and GFAP (Fig. 1f), confirming the presence of GSC-like cells along with differentiated cells. Further study of gross chromosomal rearrangements in NSG-70 cells in culture (both serum and a serum-free medium) indicated the cell line to be aneuploid with random loss and gain of various chromosomes in different metaphases, which led to a near tetraploid complement (Fig. 1g-i). The most striking feature in a large majority of cells in the population was the presence of an isochromosome of the long arm of chromosome 9 (Isochromosome 9q or i9q). Chromosomal or cytogenomic microarray analysis that involves differential binding of specific probes to the site of genomic imbalances to enable identification of deletions/loss, duplications (gain/ amplification), unbalanced translocations validated the presence of isochromosome 9q, with a specific gain of cytoband q21.11-q34.3 (Fig. 1g-ii). Strikingly, this location harbors the NOTCH1 gene, which is of special interest in the context of GSC regulation and functions.

NSG-70 GSC culture is likely to represent an in vitro model of mixed lineage
In order to understand the molecular networks regulating NSG-70 cells, we profiled their gene expression along with two other GSC lines earlier established in the lab (KW and MUR) and further compared them with the expression of NB primary cultures by RNA sequencing. We used the DAVID online analysis tool (https:// david. ncifc rf. gov/) for the functional analysis of pathways significantly enriched in the list of differentially expressed genes among the 3 GSC cultures vs. normal brain culture to identify statistically significantly enriched GO terms after uploading all downregulated genes. GO analysis results showed that downregulated DEGs were significantly enriched in microtubule binding, DNA replication origin binding, cell cycle and histone binding. In addition, upregulated DEGs were significantly enriched in the focal adhesion, calcium ion binding and extracellular matrix structural constituents (Fig.  S2b). It indicated significant alteration of specific genes and pathways in GSCs over normal, suggesting their possible contribution to gliomagenesis (Fig. S2a).
We hence oriented our study to compare the levels of stem cell and differentiation markers in the GSC lines relative to hNSC and NB cells. All cells expressed considerable heterogeneity of these markers, although high levels of CD44 expression in the 3 GSCs and lower levels of stem cell markers in NB cells were observed (NESTIN, OCT4 and NANOG; Fig. 2a, Fig. S2c), possibly indicating their roles in maintaining progenitor pool and multipotency. Significantly higher expression of oligodendrocyte lineage transcription factor 2 (OLIG2), neuronal precursor marker doublecortin (DCX), and astrocytic marker (GFAP) was identified in GSCs compared to NB; the latter suggests astrocytes to be the predominant population (Fig. 2b). OLIG2 expression was highest in NSG-70 cells, suggesting that a subset of cells is committed to an oligodendroglial fate towards a proneural phenotype. The heterogeneity in the expression of stem cell and differentiation markers was also reflected at the protein level (Fig. 2c, Fig. S2d).
Further, to delineate the subtype of GSC cultures, we profiled the expression levels of proneural and mesenchymal subtype-specific genes in the 3 GSC lines, which revealed KW to be predominantly proneural and MUR as mesenchymal, while NSG-70 presents with a mixed lineage (proneural and mesenchymal subtypes (Fig. 2d). In conclusion, NSG-70 is a new primary glioblastoma cell line representing a mixed (proneural and mesenchymal) subtype with a high expression of self-renewal molecules indicating its inherent stem-like nature with the potential to differentiate into different lineages, along with a striking feature of a novel marker i9p marker.

NOTCH1 knockdown leads to a reduction in stemness and mesenchymal markers, along with decreasing WNT5A expression
To elucidate the role of the Notch pathway in self-renewal and differentiation of stem cells, we first evaluated the effects of NOTCH1 knockdown on stemness in the three GSC lines (NSG-70, KW, and MUR). Our results demonstrated diminished expression of stemness as well as mesenchymal markers such as CD44, TWIST1, and FN1 in the 3 GSC lines (Fig. 3d-i, d-ii, g, h, Fig. S3e, S3f). The crosstalk of the Notch1 and Wnt signaling pathways has been indicated to be crucial in stem cell maintenance [27]. However, recent reports suggested the role of Wnt5a in inducing mesenchymal-like characteristics in GSCs [19]; there is a dearth of knowledge on the crosstalk of NOTCH1 and WNT5A. Hence, we sought to understand the significance of these signaling pathways in glioma. We identified significantly higher expression of NOTCH1 and WNT5A mRNA in primary GBM tumor tissues compared to normal tissues in the UALCAN database ( Fig. 3a-i, a-ii). Further examining the expression levels of these two molecules in the three GSC lines, we observed that MUR cells expressed the highest levels of both NOTCH1 and WNT5A, whereas KW and NSG-70 expressed moderate levels of NOTCH1 and very low levels of WNT5A (Fig. 3b, c). Interestingly, knockdown of WNT5A alone in WNT5A-high cells, i.e., MUR, also showed significantly decreased stem cell and mesenchymal markers (Fig. 4a, b), suggesting the coordinated action of NOTCH1 and WNT5A in maintaining stemness and promoting mesenchymal characteristics in GSCs.

WNT5A overexpression or induction in GSCs enhances NOTCH1 levels, stem cell, and mesenchymal marker expression
In either case of NOTCH1 and WNT5A knockdown, reduction in the levels of stemness and mesenchymal markers at ◂ RNA and protein levels were concurrently detected in GSC lines; this suggests a coregulation of NOTCH1 and WNT5A in GSCs. Prompted by this observation, we assessed NOTCH1 expression upon overexpression of WNT5A in NSG-70 and KW (WNT5A-low GSCs) at RNA and protein levels to explore if WNT5A could influence NOTCH1 levels. The data indicated elevated expression of NOTCH1 and its target gene HES1, stem markers (NESTIN, SOX2), and mesenchymal markers (FN1, TWIST1, VIMENTIN) in WNT5A overexpressed GSCs compared to their control counterparts (Fig. 4c, d). This suggests that components of the WNT5A pathway synergize with NOTCH1, promoting GSC characteristics and thereby contributing to tumor progression.

WNT5A knockdown drives GSCs into the proliferative phase
We also detected lower levels of p21 upon NOTCH1 knockdown in NSG-70 and MUR (Fig. 2h-ii, 2i), whereas higher p21 levels after WNT5A overexpression (Fig. 4c), which indicated the involvement of WNT5A in cellular proliferation.
We further assessed the cell cycle progression of MUR (WNT5A silenced) as well as NSG-70 and KW (WNT5A overexpressed or rhWNT5A induced). Our data demonstrated that WNT5A knockdown in MUR facilitates cell proliferation marked by an increase in S phase cells (Fig.   Fig. 2 Profiling of GSCs for stemness, differentiation, and subtypespecific markers. Heat map representing expression levels of a stem cell markers (CD133, CD44, NESTIN, SOX2, OCT4, and NANOG) and b differentiation markers (DCX, GFAP, MAP2, OLIG2, and TUJ1) in the three GSC cultures (KW, MUR, and NSG-70) and normal brain (NB) culture relative to human neural stem cell (hNSC) culture by qPCR; c Expression analysis of stem cell markers (NES-TIN, SOX2, MSI1, and CD44) along with astrocytic marker, GFAP in the three GSCs (KW, MUR, and NSG-70) and hNSC culture by western blotting; d qPCR showing expression levels of glioblastoma subtype markers-proneural (PDGFRA, NKX2-2, OLIG2, P21, and SOX2) and mesenchymal (CD44, FN1, TWIST1, YKL40, and VIM) in GSCs (NSG-70, KW, and MUR) calibrated with hNSCs. ACTB was used as the reference gene. One-way ANOVA followed by Dunnet's post hoc test was used to determine statistical significance. N = 3; ns-non-significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 S4b). Flow cytometry and immunofluorescence to evaluate the Ki67 positive population, along with MTT assay in MUR, also displayed similar results (Fig. 4e, f, and g). All these data suggested that WNT5A knockdown drives GSCs into the proliferative phase. Contrarily, the reduction in cellular proliferation upon WNT5A overexpression in WNT5A-low GSCs (NSG-70 and KW) as indicated by flow cytometry to measure the Ki67 positive population as well as MTT assay (Fig. 4h-i, h-ii, i-i, and i-ii), supports the role of WNT5A in maintaining GSC quiescence.

NOTCH1, in association with WNT5A promotes vasculogenic mimicry in GSCs besides maintaining stem cell vigor
The decreased levels of VEGFA and tumorigenicity observed in KW cells following NOTCH1 knockdown (Fig. S3h, Fig. 5a), along with NOTCH1 being an important regulator of vasculogenic mimicry (Fig. 5d), led us to evaluate whether rhWNT5A too would promote vasculogenic mimicry in the WNT5A-low GSC (NSG-70). Our data showed an almost 50% increase in the total number of junctions, as well as the tube length in rhWNT5A treated cells compared to control. NOTCH1 knockdown significantly decreased both these parameters by approximately 75%, whereas rhWNT5A treatment of NOTCH1 silenced cells effectively rescued the ability to form vessel-like structures (Fig. 5d). Additionally, WNT5A induction of GSCs also resulted in higher expression of not only NOTCH1 but also other angiogenesis-related genes, including CD31, VEGFA, and PDGFRα (Fig. 5c). This underlines the involvement of WNT5A in promoting vasculogenic mimicry in GSCs and might represent one of the crucial alternative ways by which GSCs promote tumor neovascularization.
Further, to investigate whether silencing NOTCH1 and WNT5A expression was associated with reduced stem cell vigor of GSCs, we analyzed the surface expression of the GSC marker CD44 by flow cytometry. NOTCH1 knockdown showed a significant reduction in the CD44 expression in KW and MUR (Fig. S3g, Fig. 3i). WNT5A knockdown in MUR also showed a decrease in CD44 levels (Fig. S4a). Further, to compare the frequencies of stem cells in control and NOTCH1 knockdown MUR cells, we compared their neurosphere formation ability and observed a significant reduction in the number as well as the size of neurospheres generated from single cells (Fig. 5e). Similarly, we also observed a remarkable decrease in the anchorage independent tumor cell growth and invasive ability of MUR upon NOTCH1 as well as WNT5A knockdown, as represented by the soft agar assay (Fig. 5f). All this together implies the collective role of NOTCH1 and WNT5A in maintaining stem cell characteristics in GSCs.

Discussion
GBM tumors are extremely heterogeneous, with numerous reports supporting the likely involvement of GSCs in the maintenance of cellular heterogeneity. GSCs play a crucial role in the emergence and progression of GBM tumors, substantiating GSCs as a key target for treatment. Accumulating evidence emphasizes several developmental signaling pathways such as the Notch, Wnt, and Sonic hedgehog (SHH) in tweaking the differentiation and proliferation of GSCs, along with modulating tumor invasion and neoangiogenesis [11].
Hence, in the current study, we first established and characterized a GSC culture, referred to as NSG-70, from recurrent grade IV glioma patient-derived tumor tissue to comprehend the role of GSCs in tumor sustenance. Since the characteristic property of stem cells is their ability to differentiate into various cell types, we examined the stem cell characteristics of NSG-70 GSCs by analyzing the expression of stem cell markers such as CD133, SOX2, NESTIN, and MSI1 along with differentiation markers such as GFAP, S100β, TUJ1, and MAP2 by qPCR and immunofluorescence. The data showed positivity for all the stem cell markers with very few TUJ1 + and MAP2 + neural-like cells, confirming the stemlike feature of NSG-70 culture. As GSCs are also reported as the quiescent population resistant to treatments that preferentially target rapidly dividing cells, we performed IF staining for Ki67, MTT, and PKH dye dilution assays to check the proliferative status of NSG-70 GSCs. The data showed only a few positive cells in Ki67 staining and extremely slow dilution of the PKH dye, collectively suggesting the slow dividing nature of the NSG-70 culture.
There are ambiguous beliefs regarding the origin of GSCs, with some groups supporting the idea that GSCs arise from developmentally arrested neural progenitors. In contrast, others propose that they emerge from dedifferentiated astrocytes [28]. Sandberg et al. showed that GSCs and adult human NSCs (ahNSCs) share similar levels of stemness and neural lineage markers; however, developmental pathway genes are significantly upregulated in GSCs compared to ahNSCs [29]. We used iPSC-derived hNSCs to compare the expression pattern of stem cells as well as neural markers in NSG-70 with respect to hNSCs. Here, we utilized two other previously established grade IV GSC cultures (KW and MUR) for comparison. In our study, we observed a very heterogeneous expression pattern for all these markers among the 3 GSC cultures with respect to hNSC, except for the remarkably high expression of CD44 in GSC cultures compared to hNSC. It might suggest the inherent heterogeneity among the GSC cultures specifying the need to assign tumor subtype to these cultures based on predominant transcriptional profiles and specific genetic alterations as described by Verhaak et al. [6]. GSC subtyping will ease the complexities involved in the study to use them as a model for personalized therapy and achieving a better prognosis. Verhaak et al. categorized GBM tumors into 4 distinct molecular subtypes-classic, neural, proneural, and mesenchymal [6]. However, Lottaz et al. proposed only 2 subgroups of GBM-proneural and mesenchymal, which possesses the transcriptional profile of fetal NSC (fNSC) and adult NSC (aNSC), respectively. This subgrouping highlights the different origin and activation of different signaling pathways in proneural and mesenchymal tumors [7]. Here, we profiled the 3 GSCs along with hNSC, and we found KW as proneural, MUR as mesenchymal, and NSG-70 as a mix of proneural and mesenchymal subtypes.
Since all tumors with IDH mutation are essentially secondary glioblastomas that reflect proneural subtype expression, we determined the mutation status of IDH1 by staining for mutated-IDH1 along with mutated p53 and wild type ATRX using IHC. NSG-70 xenograft sections were negative for IDH1 mutation and showed ATRX retention, confirming NSG-70 as a primary tumor, which could belong to virtually any subtype. KW xenografts, on the other hand, showed IDH1 mutation and ATRX loss, indicating KW tumor as secondary glioblastoma of proneural subtype. Additionally, glioblastomas are cytogenetically heterogeneous tumors.
Hence, we determined the various chromosomal alterations in the NSG-70 GSC line by karyotype analysis which revealed a unique cytogenomic feature, i.e., isochromosome 9q (i9q) along with several previously reported chromosomal aberrations. We then validated the isochromosome 9q presence by chromosome microarray analysis, which also illustrated a specific gain of cytoband q21.11-q34.3, harboring the NOTCH1 gene. This finding corroborated with the accumulating evidence suggesting the crucial role of Notch signaling in the NSCs and its deregulation in GBM [30], wherein activated Notch signaling is correlated with the stem-like phenotype of GSCs [31].
The Notch pathway has been demonstrated to promote proliferation and suppress differentiation of neural precursor cells [32]. Hence, we first investigated the effect of NOTCH1 on GSC stemness. Our results demonstrated that NOTCH1 knockdown significantly inhibited the stemness markers such as NESTIN and SOX2 as well as mesenchymal markers-CD44 and TWIST1, along with concomitant downregulation of WNT5A. Recent reports demonstrated WNT5A to promote epithelial-mesenchymal transition, migration, and invasion in non-small-cell lung cancer and pancreatic cancer [33,34]. In GBM tissues also, we found significantly higher mRNA expression of NOTCH1 and WNT5A compared to the normal brain according to the TCGA data-based UAL-CAN database. This observation was supported by the high RNA and protein levels of NOTCH1 in all the 3 GSCs; however, they showed very different levels of WNT5A. WNT5A was expressed highly in the mesenchymal subtype of GSC (MUR), whereas in proneural (KW) and proneural-mesenchymal mix (NSG-70) GSCs, WNT5A levels were low at RNA and protein levels. Given the fact that the NSG-70 line was a recurrent GBM-derived line, phenotypic plasticity of GSCs wherein proneural to mesenchymal subtype transition or vice-versa occurs, resulting in therapy resistance and tumor recurrence [35], might be the cause for mixed subtype in NSG-70. It also signifies the subtype-specific expression of WNT5A in GSCs. This is in line with the previous studies showing WNT5A overexpression induces infiltrative, mesenchymal-like characteristics in high-grade gliomas [19].
Notch and Wnt/β-catenin pathways modulate motility and invasion by activating an epithelial-mesenchymal-like program in GBM [10]. However, any elaborate study showing the role of non-canonical WNT5A in the regulation of the maintenance and proliferation of GSCs is lacking. Hence, we next examined the effect of WNT5A downregulation in MUR (WNT5A-high GSC), and we found a Fig. 3 Knockdown of NOTCH1 reduces stem cell markers and promotes proliferation of GSCs a-I, a-ii Expression analysis of NOTCH1 and WNT5A in GBM patients using UALCAN database; Analysis of endogenous level expression of NOTCH1 and WNT5A in 3 GSC cultures (KW, MUR, and NSG-70) by b qPCR and c western blotting; d-i, d-ii Expression analysis of NOTCH1, HES1, NESTIN, WNT5A, and mesenchymal genes (CD44 and TWIST1) upon stable knockdown of NOTCH1 gene in NSG-70 GSC culture by qPCR and western blotting; e MTT assay to determine proliferative ability of NSG-70 GSC upon NOTCH1 Knockdown (N = 3); f Immunofluorescence analysis of Ki67 positive cells in NSG-70 upon NOTCH1 knockdown, followed by its graphical representation. The graph represents mean ± SD calculated from 4 different fields in the experiment performed in triplicates (Scale bar: 50 μm, magnification: 20X); g Expression analysis of NOTCH1, HES1, NESTIN, SOX2, WNT5A, TWIST1 and FN1 by qPCR in MUR GSCs upon stable knockdown of NOTCH1 gene; h Expression analysis of NOTCH1, NESTIN, NANOG, OCT4, and mesenchymal genes (CD44 and TWIST1) upon stable knockdown of NOTCH1 by western blotting in MUR GSC culture; i Flow cytometric analysis of CD44 positive population in MUR GSC upon NOTCH1 knockdown. Paired student's t-test was used to determine statistical significance. N = 3; ns-non-significant; *p < 0.05; **p < 0.01; ***p < 0.001 ◂ remarkable reduction in stem cell as well as mesenchymal markers, along with NOTCH1 inhibition at RNA and protein levels.
One of the hallmarks of GBM is rapid infiltrating growth and extensive vasculature. Several reports elicited induction of VEGFA and Notch signaling genes as an angiogenic response to hypoxia. Our data showed reduced VEGFA level (Fig. S3h) and vasculogenic mimicry in GSCs (NSG-70 and MUR) upon NOTCH1 knockdown. Consistent with the previous findings that Notch inhibition leads to the reduced tumorigenicity of GSCs [36], we observed reduced tumor volume with very less vascularity in NOTCH1 knockdown KW injected mice compared to control. Representative KW xenografts stained for CD34 (a direct measure of the tumor vessel density) demonstrated a significantly enhanced accumulation of CD34 + cells in control tumors compared to NOTCH1 knockdown KW-derived tumors (Fig. 5b). This suggests that NOTCH1 might play a role in promoting tumor angiogenesis.
Moreover, the role of Wnt5a in inducing endothelial cell proliferation and survival under serum-deprived conditions had been documented earlier [17]. Recently, GSC differentiation into endothelial-like cells (GdECs) mediated by WNT5A was shown by Hu B et al. [18]. However, there are no studies showing the involvement of WNT5A in vasculogenic mimicry of GSCs to date in glioma or the coordination of WNT5A and NOTCH1 in GSC maintenance. Our results showed that NOTCH1 knockdown lowers the expression of angiogenic molecules as well as vasculogenic mimicry in GSCs. On the contrary, upon rhWNT5A induction of GSCs, we noted an increase in the levels of angiogenic molecules as well as vessel branches representative of vasculogenic mimicry (Fig. 5c, d). Together, our results reveal the collective role of NOTCH1 and WNT5A signaling in tumor angiogenesis.
Current work elucidates the coordinated action of NOTCH1 and WNT5A signaling pathways in GSC biology. The work focuses on improving the understanding of GSC self-renewal and maintenance and how NOTCH1 and WNT5A signaling are closely associated with modulating tumor angiogenesis. From this work, it seems clear that the effects of NOTCH1 and WNT5A are in sync, but further investigation is required to shed light on whether the coinciding effect of the two pathways is mediated by NOTCH1 itself or downstream effectors of the pathway. However, both the molecules cooperate to modulate stem cell self-renewal, proliferation, and angiogenesis in GSCs (Fig. 6). We believe that the NOTCH1-WNT5A axis could be harnessed in the Fig. 4 Stable knockdown of WNT5A suppresses stem cell and mesenchymal markers and drives GSCs into proliferative phase a Expression analysis of NOTCH1, NESTIN, and mesenchymal markers (CD44, TWIST1, and FN1) upon knockdown of WNT5A by qPCR in MUR (WNT5A-high) GSC; b Expression analysis of stem cell (NESTIN, OCT4, and NANOG), and mesenchymal markers (CD44, TWIST1, and FN1), and P21 upon knockdown of WNT5A by western blotting in MUR GSC; c Expression analysis of stem cell and mesenchymal markers upon WNT5A overexpression in WNT5Alow GSCs c NSG-70 and d KW; e MTT assay to determine proliferative ability of MUR GSC upon WNT5A knockdown; f Immunofluorescence analysis of Ki67 positive cells in MUR upon WNT5A knockdown, followed by its graphical representation. (N = 3) Paired student's t-test was used to determine statistical significance.; g Flow cytometric analysis of Ki67 positive population in MUR upon WNT5A knockdown, followed by its graphical representation; MTT assay to determine the proliferative ability of h-i NSG-70 and h-ii KW GSC upon WNT5A overexpression; Flow cytometric analysis of Ki67 positive population in WNT5A low-GSCs i-i KW and i-ii NSG-70 upon WNT5A overexpression. N = 3; Paired student's t-test was used to determine statistical significance; *p < 0.05; **p < 0.01; ***p < 0.001 ◂ future as a promising link to develop effective GSC-directed therapies in GBM.
In conclusion, NOTCH1 and WNT5A signaling pathways work synergistically and regulate GSCs self-renewal, proliferation, invasion, and vasculogenic mimicry to aid tumor neoangiogenesis. NOTCH1-WNT5A axis needs to be further explored, which could have great clinical implications and yield promising therapeutic targets for GBM. association with WNT5A promotes vasculogenic mimicry besides maintaining stem cell vigor in GSCs a Control and NOTCH1 knockdown KW GSC-derived tumor in NOD-SCID mice, followed by graphical representation of tumor volume; b Representative immunohistochemistry images of control and NOTCH1 knockdown KW xenograft sections stained for CD34. (Scale bar: 100 μm, magnification: 10X); c qPCR analysis of angiogenic gene expression in control and NOTCH1 knockdown NSG-70 upon rhWNT5A induction; d Matrigel tube formation assay to assess the effect of rhWNT5A on vasculogenic mimicry in NSG-70 GSCs followed by its graphical representation; e Neurosphere formation assay to determine self-renewal potential of control, NOTCH1 and WNT5A knockdown MUR, followed by its graphical representation; f Soft agar colony formation assay to determine the invasive potential of control, NOTCH1 and WNT5A knockdown MUR, followed by its graphical representation; g Soft agar colony formation assay to determine the invasive potential of control and NOTCH1 knockdown NSG-70, followed by its graphical representation Paired student's t-test was used to determine statistical significance. N = 3; ns = non-significant; *p < 0.05; **p < 0.01; ***p < 0.001 ◂ Fig. 6 A schematic representation showing the role of NOTCH1 and WNT5A coregulation in modulating stem cell characteristics and vasculogenic mimicry. GSCs belonging to the mesenchymal subtype express high levels of NOTCH1 and WNT5A. Knockdown of either NOTCH1 or WNT5A results in suppressed stem cell markers such as NESTIN, SOX2, and CD44 and self-renewal property. Reduced levels of NOTCH1 and WNT5A drive GSCs towards proliferation and stem cell exhaustion along with negatively regulating vasculogenic mimicry potential