Transforming Growth Factor-Beta-Regulated LncRNA-MUF Promotes Invasion by Modulating the miR-34a Snail1 Axis in Glioblastoma Multiforme

Transforming growth factor beta (TGF-β)-regulated long-non-coding RNAs (lncRNAs) modulate several aspects of tumor development such as proliferation, invasion, metastasis, epithelial to mesenchymal transition (EMT), and drug resistance in various cancers, including Glioblastoma multiforme (GBM). We identified several novel differentially expressed lncRNAs upon TGF-β treatment in glioma cells using genome-wide microarray screening. We show that TGF-β induces lncRNA-MUF in glioma cells, and its expression is significantly upregulated in glioma tissues and is associated with poor overall survival of GBM patients. Knockdown of lncRNA-MUF reduces proliferation, migration, and invasion in glioma cells and sensitizes them to temozolomide (TMZ)-induced apoptosis. In addition, lncRNA-MUF downregulation impairs TGF-β-induced smad2/3 phosphorylation. In line with its role in regulating invasion, lncRNA-MUF functions as a competing endogenous RNA (ceRNA) for miR-34a and promotes Snail1 expression. Collectively, our findings suggest lncRNA-MUF as an attractive therapeutic target for GBM.


INTRODUCTION
GBM is a heterogeneous malignancy of the central nervous system characterized by aggressive invasion into the surrounding tissue (1). Despite following an aggressive treatment approach involving surgical resection and radio-and chemotherapy with temozolomide (TMZ), it remains incurable with a dismal survival rate of about 15 months (2). One of the characteristic features of GBM is extensive infiltration and invasion of the tumor cells to the surrounding parenchyma, which leads to colonization and relapse of tumors (3). TGF-b is a cytokine with multiple functions regulating cell proliferation, differentiation, and tissue homeostasis (4,5). TGF-b promotes cancer cell invasion, EMT, and chemoresistance (5). TGF-b is overexpressed in glioblastoma, and its elevated expression is associated with the increased histologic grade of GBM (6). TGF-b promotes proliferation, invasion, metastasis, angiogenesis, resistance to apoptosis, replicative immortality, evasion of growth suppression, evasion of immune checkpoint blockade, and chemoresistance in GBM (7)(8)(9).
Using a microarray screen, we identified several previously uncharacterized TGF-b1-regulated lncRNAs in T98G cells and characterized the role of one of the TGF-b-induced lncRNAmesenchymal upregulated factor (lncRNA-MUF/LINC00941) in glioma physiology. LncRNA-MUF was first identified by Yan et al., and they demonstrated that it regulates EMT in hepatocellular carcinoma (HCC) (26). However, the functions and mechanism of action of lncRNA-MUF in GBM were not known. We show that levels of lncRNA-MUF are upregulated in GBM tumor samples along with histological grade. Our results suggest that it functions as an oncogenic lncRNA to promote glioma cell growth and invasion by functioning as a miRNA sponge for miR-34a that targets and suppresses Snail1. In addition, we show that lncRNA-MUF depletion sensitizes glioma cells to TMZ-induced apoptosis. Collectively, our results suggest that the lncRNA-MUF/miR-34a/Snail1 signaling axis may serve as a novel therapeutic target for GBM treatment.

Cell Culture and Treatments
T98G cells were purchased from the American Type Culture Collection (Manassas, VA). LN229, LN18, and U87-MG cells were purchased from NCCS, Pune. All cells were grown in complete medium, DMEM (Invitrogen) containing 10% fetal bovine serum (FBS) supplemented with 1 mM l-glutamine, and penicillin/streptomycin (Gibco). Cells were treated with TGF-b1 (10 ng/ml) PeproTech (#100-21) in serum-free medium (SFM) for dose and duration indicated in the figures and legends. SB505124 (Tocris # 3263), an inhibitor of TGFbRI/smad2/3, was used at a concentration of 6 µM for pretreatment of GBM cells to inhibit TGF-b signaling wherever indicated.

Microarray Analysis
Agilent SurePrint G3 Gene Expression Microarrays for Human (v3) for lncRNAs, containing 30,606 lncRNAs and 37,756 RefSeq-coding transcripts, were used to interrogate lncRNA and mRNA changes in vehicle versus TGF-b1 (10 ng/ml)treated T98G cells after 24 h. RNA was isolated using MN NucleoSpin RNA Plus isolation kit (Cat. No. 740984.5). 10 µg of purified RNA samples was treated with recombinant DNAse I (Invitrogen Thermo Scientific-Cat. No. EN0521) as per manufacturer's instructions, and the RNA samples were column purified using the MN NucleoSpin column purification kit. Hybridization and analysis were performed at the Molecular Genomics Core at Genotypic Technology (Bangalore). Briefly, total RNA was end-labeled using Agilent Quick-Amp labeling Kit (p/n5190-0442) and hybridized to Agilent Human Gene Expression Microarray 8X60K. Fragmentation of labeled cRNA and hybridization were done using the Gene Expression Hybridization kit (Agilent Technologies, In situ Hybridization Kit, # 5190-0404). The hybridized slides were scanned using the Agilent Microarray Scanner (Agilent Technologies, Part Number G2600D). Data analysis was done by using GeneSpring GX software version 14.5. Gene expression in the test group (TGFb) was compared with the control group (C) to identify differentially expressed genes (DEGs) upon TGF-b treatment. DEGs were selected based on log base 2 (fold ≥ 0.6) and log base 2 (fold ≤-0.6) with a statistical significance of p-value < 0.05.

RNA Isolation and Real-Time PCR
Total RNA was extracted from glioma cells using the MN NucleoSpin RNA plus isolation kit (Cat. No.: 740984.5). 1 µg of RNA was converted into cDNA using the PrimeScript firststrand cDNA kit from Takara (#6110A). Quantitative real-time PCR (qRT-PCR) was performed with the SYBR Green PCR Kit (#RR820A, Takara) in the Bio-Rad CFX96 real-time qPCR system. All reactions were performed in triplicates and normalized with TBP/HPRT as an internal control. The relative gene expression of each sample was calculated using the 2 -ddct formula. For miRNA expression analysis, RNA was isolated using Zymo Quick-RNA ™ Miniprep Plus Kit (#R1057). miRNA cDNA was synthesized using the mir-X miRNA 1 st -Strand Synthesis Kit (#638313, Takara). qRT-PCR of miRNA was carried out using the universal primer and the primer specific to miR-34a-5p (Supplementary Table 1

Nuclear and Cytoplasmic Extract Preparations
Cytoplasmic and nuclear fractions were separated, and RNA was purified as described previously (27). qRT-PCR was performed to identify relative RNA levels in each fraction by using GAPDH as a control for cytoplasmic fraction, and MALAT1 as a control for nuclear fraction.

Cell Proliferation Assay
Colorimetric cell proliferation assay was performed by using the WST-1 reagent (Cat#. 05015944001, Roche) at the indicated time according to the manufacturer's instructions. Briefly, cells were seeded at a concentration of 2,500-5,000 cells/well in 96-well plates and transfected with siRNAs si-MUF-1, si-MUF-2, and si-NS at 40 nM, and cell proliferation was quantified at OD of 450 nm.

Colony Formation Assay
For clonogenic assays, cells were seeded into 96-well dishes and treated with si-NS or siRNAs against lncRNA-MUF. 24 h posttransfection, cells were trypsinized and seeded at a density of 200 cells per well in a 6-well dish and incubated at 37°C. Media were changed every 3 days. Colonies formed 14 days after plating were fixed with 4% paraformaldehyde and stained with crystal violet solution, and counted.

Caspase 3/7 Assay
Luminometric assay kit for caspase-3/7 (Promega, G8090) was used to determine the enzymatic activity of caspase-3/7 in glioma cells transfected with si-MUF-1 and 2. 48 h post-transfection, proluciferin DEVD substrate and caspase-Glo 3/7 buffer were added to the cells, and the assay was performed as per the manufacturer's instructions.

Migration Assays
For migration assay, si-MUF-1/si-MUF-2 or negative control siRNA-transfected glioma cells were seeded in a 12-well dish and cultured overnight. Scratch was made using a 20-µl pipette tip followed by PBS wash. Cells were maintained in 0.5% serumcontaining media. Images of scratch were taken at 0, 24, and 48 h, and the migrating length was calculated using ImageJ (30).
For lncRNA-MUF promoter analysis, we cloned the -734-bp promoter region of lncRNA-MUF with predicted putative SBEs into the restriction sites of SacI and NheI of pGL3basic luciferase and renilla_polyA construct (a gift from Oskar Laur) (Addgene plasmid # 128046; http://n2t.net/addgene:128046; RRID: Addgene_128046). T98G and U87-MG cells were seeded at~60%-70% confluency in 24-well plates. The next day, they were transiently transfected with 0.3 mg of lncRNA-MUF promoter containing pGL3basic luciferase and renilla_polyA reporter plasmid using jet prime transfection reagent. Eighteen hours post-transfection, cells were serum-starved for 6 h followed by treatment with 10 ng/ml TGF-b1 for the indicated time. Luciferase activity was measured using the Dual-Luciferase ® Reporter Assay System according to the manufacturer's protocol (Promega) on the SpectraMax iD3 Luminometer (Molecular Devices Corporation). The results are expressed as a fold change in luciferase activity over control (28).

Statistical Analysis
Results are presented as mean ± SEM unless otherwise stated. We used paired Student's t-test for comparisons between two experimental groups. Additional statistical tests information is described in the figure legends. p < 0.05 was considered statistically significant.

Identification of TGF-b-Regulated LncRNAs in GBM Cells Using Microarray Screen
We sought to identify, in an unbiased fashion and at a genomewide scale, differentially expressed lncRNAs upon TGF-b treatment in glioma cells. We performed gene expression analysis of control, and TGF-b1-treated T98G glioma cells using the Agilent SurePrint G3 Gene Expression Microarrays for Human (v3) for lncRNAs. Using a 1.5-fold change and pvalue < 0.05 as a threshold, we identified 91 differentially expressed lncRNAs and 397 differentially expressed mRNAs in our screen ( Figures 1A, B and Supplementary Table 2). LncRNAs constitute 18.3% of transcripts among the total number of DEGs identified upon TGF-b1 treatment in T98G cells ( Figure 1B). We verified the TGF-b1-induced gene expression changes in levels of lncRNAs in T98G cells using qRT-PCR ( Figure 1C). LncRNAs ENST00000409910 and LOC79160 get~4-fold upregulated upon TGF-b treatment ( Figure 1C). LncRNAs LINC00312, LOC101928710, lncRNA-MUF, and lnc-EGR2-1 get~1.5-3-fold upregulated upon TGF-b treatment ( Figure 1C). LncRNAs CTB-178M22.2 and KCNMA1-AS1 are significantly downregulated upon TGF-b treatment ( Figure 1C). The expression of several TGF-bregulated mRNAs identified from the microarray screen was also verified using qRT-PCR ( Figure S1A). Among these upregulated lncRNAs, we further set out to characterize the role of lncRNA-MUF in glioma pathogenesis.

LncRNA-MUF Is Upregulated in GBM Tumor Samples and Is Associated With Poor Patient Prognosis
To investigate the role of lncRNA-MUF in GBM pathophysiology, we decided to evaluate its expression in GBM tumor samples using the CGGA database (http://www.cgga.org.cn/). Using the mRNAseq_693 dataset of the CGGA database, we found that levels of lncRNA-MUF were significantly higher in GBM samples than normal brain tissues (p = 6.3e-15 ) ( Figure S1B). Moreover, lncRNA-MUF levels are significantly higher in grade IV GBM than in lower-grade gliomas (p = 1.6e-17 ) ( Figure 1D). GBM patients with IDH mutation show a better survival rate than the IDH wild-type group (1). Hence, we evaluated the expression of lncRNA-MUF in IDH mutant and wild-type glioma samples. We observed that the expression of lncRNA-MUF is significantly higher in gliomas with the IDH wild-type group than in the IDH mutant group (p = 9.7e-28 ) ( Figure 1E). In addition, high expression of lncRNA-MUF is correlated with poor overall survival in both primary and recurrent GBM patients (p = 0.0063) ( Figure 1F). These results suggest that lncRNA-MUF expression is associated with aggressive phenotype and poor survival in glioma patients. To this end, GBM cells were treated with 6 µm SB505124 (TGFbR1/smad2/3 inhibitor) for 2 h before treatment with TGF-b1 (24 h). Blocking smad2/3 with SB505124 significantly abrogated TGFb-induced lncRNA-MUF expression in glioma cells (~50% reduction in T98G, LN229, and U87-MG) ( Figure 2C). As TGF-b-induced lncRNA-MUF expression and TGF-bRI inhibitor significantly abrogated lncRNA-MUF levels, we used luciferase reporter assay to confirm if lncRNA-MUF promoter can drive TGF-b-mediated luciferase activity. Transfection of T98G and U87-MG cells with the lncRNA-MUF-promoterluciferase reporter construct followed by TGF-b treatment for 24 h resulted in a significant~2.5-and 2-fold increase in luciferase activity over control, respectively ( Figure S2C). Next, we performed ChIP-qPCR to determine whether TGF-b promotes increased binding of smad2/3 to SBE on the lncRNA-MUF promoter. ChIP-qPCR revealed increased binding of smad2/3 on SBE on the lncRNA-MUF promoter upon TGF-b stimulation ( Figure 2D). These results suggest that TGF-b upregulates lncRNA-MUF expression through the canonical SMAD signaling pathway.

Knockdown of LncRNA-MUF Reduces Cell Proliferation, Induces Apoptosis, and Sensitizes Glioma Cells to TMZ-Mediated Apoptosis
To investigate the physiological function of lncRNA-MUF in glioma pathogenesis, we established lncRNA-MUF knockdown by siRNA using two different siRNAs (si-MUF-1 and si-MUF-2) in T98G and U87-MG cell lines. The knockdown of lncRNA-MUF with si-MUF-1 results in~85% reduction, and si-MUF-2 results in~67% reduction of lncRNA-MUF levels in T98G and U87-MG cells ( Figure S3). LncRNA-MUF depletion using siRNAs results in a time-dependent reduction in cell , and U87-MG) were pretreated with 6 µM of SB505124 (TGFbR1/Smad2/3 inhibitor) for 2 h followed by co-treatment with TGF-b1 (10 ng/ml) for 24 h, and lncRNA-MUF transcript levels were determined by qRT-PCR. (D) ChIP-qPCR analysis of smad2/3 interaction with SBE in the lncRNA-MUF promoter in control and TGF-b-treated T98G cells. DNA was isolated from control and TGF-b-treated cells after immunoprecipitation with the anti-smad2/3 antibody and was amplified using specific primer sets. LncRNA-MUF promoter levels in immunoprecipitated samples were measured by qRT-PCR analysis, normalized to input, and represented as "fold enrichment relative to control IgG I.P." Values represent mean ± S.D. from two independent experiments. *Significant change compared to IgG (p < 0.05). #Significant change compared to control Smad2/3 (p < 0.05). Data information: RNA samples were analyzed by quantitative RT-PCR, normalized with TBP/HPRT. Error bars represent the mean ± SEM from 3 independent experiments. *Significant change compared to respective control samples (p < 0.05). # Significant decrease from TGF-b-treated cells (p < 0.05). Statistical comparisons were made using Student's t-test. proliferation in glioma cells. Cell proliferation was reduced by~40% and~55% at 48 and 72 h post-lncRNA-MUF knockdown, respectively, in T98G cells ( Figure 3A). A similar 40%-50% reduction in cell proliferation was observed in LN229 and U87-MG glioma cells transfected with siRNA against lncRNA-MUF compared to cells transfected with non-specific siRNA (si-NS) ( Figure 3A). Consistent with the reduction in cell proliferation upon lncRNA-MUF depletion, MUF knockdown resulted in a significant decrease in colony formation of~62% and 70%, respectively, in T98G and U87-MG cells compared to respective control cells transfected with si-NS ( Figure 3B and Figure S4C). Moreover, depletion of lncRNA-MUF by siRNA also results in apoptosis as demonstrated by an increase of~1.75fold, 3.6-fold, and 3.4-fold caspase 3/7 activity in T98G, U87-MG, and LN229, respectively, as compared to control cells ( Figure 3C). Consistently, the levels of caspase 9 mRNA werẽ 2-fold increased following lncRNA-MUF knockdown in T98G and U87-MG cells ( Figure S4A). TMZ is an oral alkylating drug that is used to treat GBM; however, 50% of GBM cases develop resistance to TMZ. Several GBM cell lines such as T98G and  LN229 show resistance to TMZ, and TGF-b-induced lncRNAs are known to promote TMZ resistance (9,31). Therefore, we evaluated the effect of lncRNA-MUF knockdown on TMZ sensitivity in T98G and LN229 cells. LncRNA-MUF depletion with low siRNA levels (20 nM) resulted in significantly reduced cell proliferation in TMZ-treated T98G and LN229 cells compared to si-NS-transfected cells treated with TMZ ( Figure 3D). In addition, TMZ treatment in lncRNA-MUF knockdown resulted in a significantly higher increase in caspase 3/7 activity (~5-fold) compared to si-NS-transfected T98G and LN229 cells treated with TMZ ( Figure 3E). These results suggest that lncRNA-MUF knockdown sensitizes glioma cells to TMZ-induced apoptosis. We then investigated the effect of lncRNA-MUF knockdown on glioma cell migration and invasion. Wound healing assay revealed that lncRNA-MUFdepleted T98G and U87-MG cells show~63% and~58% reduction in cell migration, respectively, compared to control cells ( Figures 3F and S4B). Matrigel invasion assay during lncRNA-MUF depletion results in~55% and~70% inhibition of cell invasion in T98G and U87-MG cells, respectively, compared to control cells ( Figure 3G). Thus, collectively these results suggest that lncRNA-MUF serves as an oncogene to promote proliferation, drug resistance, migration, and invasion in GBM cells, and targeting lncRNA-MUF is an attractive therapeutic strategy for GBM.

LncRNA-MUF Regulates Gene Expression of a Subset of TGF-b Target Genes in cis and trans
LncRNA transcripts often regulate gene expression in cis and trans (10). We first evaluated the effect of lncRNA-MUF knockdown on its cis genes ( Figure 4A). We observed~50% downregulation of the Caprin2 gene in T98G and U87-MG upon lncRNA-MUF knockdown with both the siRNAs ( Figure 4A). This is consistent with Ai et al., who reported the cis-regulation of the Caprin2 gene by lncRNA-MUF through chromosome looping in OSCC (32). However, the levels of other cis genes (IPO8, LOC645485, LOC107984476) remained unchanged upon lncRNA-MUF knockdown ( Figure 4A). We also observed that the Caprin2 gene is upregulated by TGF-b in T98G GBM cells (1.7-fold) using qPCR assays ( Figure S5A). These results suggest that lncRNA-MUF regulates TGF-b-induced expression of the Caprin2 gene in cis in glioma cells. Since lncRNA-MUF regulates genes involved in the WNT/b-catenin pathway (26,32), we evaluated the impact of lncRNA-MUF knockdown on the WNT/b-catenin pathway genes in glioma cells. Surprisingly, we did not observe any significant change in their expression upon MUF depletion in glioma cells ( Figure S5B). To further identify the genes regulated in trans by lncRNA-MUF in glioma cells, we evaluated the expression of the TGF-b gene ontology group upon its siRNA-mediated knockdown. Depleting MUF resulted in~50% downregulation of Snail1,~40% downregulation of vimentin,~60% downregulation of CTGF, and~30% downregulation of c-Myc in T98G cells and U87-MG cells ( Figure 4B). Several other TGF-b-regulated genes did not show any change in expression with MUF knockdown ( Figure S5C).
Since lncRNA-MUF depletion inhibits invasion and Snail1 regulates EMT and invasion, we evaluated EMT marker expression upon lncRNA-MUF inhibition by Western blotting.

Knockdown of LncRNA-MUF Attenuates TGF-b Signaling
TGF-b-induced lncRNAs are known to regulate the TGF-b signaling pathway via an autocrine signaling loop (33). Hence, we asked if lncRNA-MUF is also involved in regulating TGF-b signaling. To test this, we evaluated the impact of lncRNA-MUF knockdown on TGF-b-induced phosphorylation of smad2/3. Silencing lncRNA-MUF in T98G results in a~35% decrease in smad2/3 phosphorylation at 15 min and 30 min post-TGF-b treatment compared to si-NS cells treated with TGF-b. A similar reduction of~30% is observed in p-smad2/3 levels upon TGF-b treatment in U87-MG cells compared to TGF-b1-treated si-NS cells ( Figures 5A and S6). This is consistent with the fact that pathway analysis by the lncACTdb database suggests that the TGF-b signaling pathway is among the top 10 enriched signaling pathways regulated by lncRNA-MUF ( Figure 5B). Our results indicate that lncRNA-MUF regulates smad2/3 phosphorylation downstream of the TGF-b pathway in glioma cells.

LncRNA-MUF Modulates TGF-b-Induced Invasion in Glioma via
the miR-34a-5p/Snail1 Axis LncRNAs function as endogenous miRNA sponges and participate in the ceRNA regulatory network (34,35). Yan et al. have reported the direct binding of lncRNA-MUF and miR-34a using RNA immunoprecipitation (RIP) and RNA pull-down assays (26). In addition, they show that lncRNA-MUF regulates Snail1 expression by sponging miR-34a to modulate EMT in HCC cells (26). Using RNAhybrid and IntaRNA databases, we identified putative miR-34a-binding sites in lncRNA-MUF ( Figure S7A). To identify the interaction between lncRNA-MUF and miR-34a-5p, we cloned the region of lncRNA-MUF with the miR-34a-binding sites into the pmirGLO vector downstream of the firefly luciferase gene. Co-transfection with the pmirGLO-lncRNA-MUF reporter plasmid and miR-34a mimics reduced the reporter activity significantly (~70%) compared to the control cells ( Figure 6D). Since miR-34a has a well-established tumor-suppressor role in several cancers, including GBM (36), we first evaluated its expression in glioblastoma tissue using the CGGA dataset. Expression of miR-34a is lowest in grade IV GBM (p = 0.0038) ( Figure 6A). In addition, the Kaplan-Meier survival curve demonstrates that high expression of miR-34a positively correlates with better survival of glioma patients (p = 0.016) ( Figure 6B). To understand the impact of miR-34a on lncRNA-MUF regulation, we first determined its levels upon miR-34a overexpression using miRNA mimics. We observed a significant 40%-50% reduction in lncRNA-MUF expression in T98G and U87-MG cells upon treatment with miR-34a mimic ( Figure 6C). Snail1 is a well-known target of miR-34a; consistent with this, we observed that transfection of miR-34a mimics in T98G and U87-MG glioma cells significantly reduced Snail1 protein levels and knockdown of miR-34a using miRNA inhibitors reversed this effect ( Figure 6E and Figure S7B). Given that miR-34a targets lncRNA-MUF and Snail1 expression and because we observed downregulation of Snail1 upon lncRNA-MUF depletion, we explored if lncRNA-MUF could act as a ceRNA to sponge miR-34a for stabilizing Snail1 to regulate invasion in glioma cells. Invasion analysis revealed that reduction in invasion upon lncRNA-MUF knockdown is significantly reversed upon cotransfection with the miR-34a inhibitor ( Figures 6G, H). Moreover, in accordance with the role of miR-34a in the regulation of invasion by Snail1, we observed that the miR-34a inhibitor significantly restores Snail1 downregulation caused by lncRNA-MUF depletion ( Figure 6F and Figure S7C). These experiments indicate that TGF-b induced lncRNA-MUF sponges miR-34a to promote Snail1-induced invasion ( Figures 6G, H).
ChIP-seq revealed that the lncRNA-MUF promoter upon TGFb stimulation accumulates, activating H3K27ac marks (38). LncRNA-MUF induction by TGF-b in CRC cells is abrogated upon treatment with disitertide, an inhibitor of TGFbR1 (39). In line with these findings, we show that lncRNA-MUF induction by TGF-b is completely abrogated upon treatment with TGFbR1/ smad 2/3 inhibitor SB505124 in glioma cells ( Figure 2C). TGF-bregulated lncRNA-MIR100HG regulates smad2/3 phosphorylation in prostate carcinoma (33). LncRNA-MUF also regulates the TGFb signaling by preventing the SMAD4 degradation by competing with b-TrCP in CRC (39). We demonstrate for the first time that MUF downregulation attenuates TGF-b-induced phosphorylation of smad 2/3 in glioma cells. LncRNA-MUF promotes OSCC progression by mediating chromosome looping to the promoter of its cis gene, Caprin2, to activate the WNT/b-catenin signalingmediated progression of OSCC (32). Although we observed a significant downregulation of the Caprin2 gene with lncRNA-MUF knockdown, we did not observe any change in the WNT/b-  catenin signaling genes. Our results suggest that apart from regulating Caprin2 expression, lncRNA-MUF modulates the expression of several genes involved in the TGF-b pathway in glioma cells (vimentin, CTGF, c-Myc, and Snail1) with Snail1 as one of the primary targets. However, the mechanism of regulation of vimentin, N-cadherin, CTGF, and c-Myc by lncRNA-MUF needs further investigation. LncRNAs act as endogenous miRNA sponges for binding to miRNAs or participating in the ceRNA regulatory network (35). The cross talk between miRNAs and TGF-b-induced lncRNAs regulates the EMT and tumor invasion in glioma (23,25). miR-34a suppresses the proliferation and invasion in glioma (47). It is downregulated in human glioma tumors as compared to normal brain tissue (47). miR-34a has a potential tumor-suppressor role in glioma by targeting several oncogenes and also induces differentiation of glioma stem cells (47). Dai et al. recently reported that LINC00665 sponges miR-34a, which targets the angiotensin II receptor type I (AGTR1) gene to impede glioma malignancy (48). Several studies have reported that Snail1 is a direct target of miR-34a (36,49,50). Snail1 is a crucial transcription factor that promotes tumor cell invasion and EMT (51). Snail1 is often upregulated in glioma, and high expression of Snail1 is associated with poor survival of glioma patients (52). We observed a positive correlation between MUF and Snail1 expression in GBM tumor samples ( Figure S7D). We also show that lncRNA-MUF depletion in glioma cells results in reduced migration and invasion, and lncRNA-MUF promotes GBM invasion by acting as an endogenous sponge for miR-34a and causing stabilization of its target Snail1 ( Figure 6G). In addition, we show that loss of lncRNA-MUF expression reduces cell proliferation, induces apoptosis, and sensitizes glioma cells to TMZ-induced cell death. Our findings suggest that the TGF-b-regulated lncRNA-MUF/miR-34a/Snail1 signaling axis is a critical regulator of invasion in GBM (Figure 7). Our results warrant further preclinical studies on lncRNA-MUF using low-passage glioma patient-derived cell models, glioma stem cells, and in vivo models to firmly establish its role as a therapeutic target for GBM.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the following: https:// www.ncbi.nlm.nih.gov/, GSE183211.