Sp1-Induced Upregulation of LBX2-AS1 Aggravates The Progression of Glioblastoma By Targeting The miR-491-5p/LIF Axis

Background: Mounting evidences have shown the importance of lncRNAs in tumorigenesis and cancer progression. LBX2-AS1 is an oncogenic lncRNA that has been found abnormally expressed in gastric cancer and lung cancer samples. Nevertheless, the biological function of LBX2-AS1 in glioblastoma (GBM) and potential molecular mechanism are largely unclear. Methods: Relative levels of LBX2-AS1 in GBM samples and cell lines were detected by qRT-PCR and FISH. In vivo and in vitro regulatory effects of LBX2-AS1 on cell proliferation, epithelial-to-mesenchymal transition (EMT) and angiogenesis in GBM were examined through xenograft models and functional experiments, respectively. The interaction between Sp1 and LBX2-AS1 was assessed by ChIP. Through bioinformatic analyses, dual-luciferase reporter assay, RIP and Western blot, the regulation of LBX2-AS1 and miR-491-5p on the target gene leukemia Inhibitory factor (LIF) was identi�ed. Results: LBX2-AS1 was upregulated in GBM samples and cell lines, and its transcription was promoted by binding to the transcription factor Sp1. As a lncRNA mainly distributed in the cytoplasm, LBX2-AS1 upregulated LIF, and activated the LIF/STAT3 signaling by exerting the miRNA sponge effect on miR-491-5p, thus promoting cell proliferation, EMT and angiogenesis in GBM. Besides, LBX2-AS1 was unfavorable to the progression of glioma and the survival. Conclusion: Upregulated by Sp1, LBX2-AS1 promotes the progression of GBM by targeting the miR-491-5p/LIF axis. It is suggested that LBX2-AS1 may


Background
Glioblastoma multiforme (GBM, WHO IV) is the most lethal type of glioma (~50%), which mainly affects people older than 50 years and featured by local invasion and neovascularization [1][2][3].Although therapeutic progression has been made on GBM, including surgery, radiotherapy, chemotherapy, and targeted therapy, its overall prognosis is extremely poor with the 5-year survival of lower than 10% [4][5][6].It is well known that the generation and progression of glioma is complicated, involving gene mutations, epithelial-mesenchymal transition (EMT), excessive angiogenesis and other biological processes [7,8].
So, a clear understanding of the molecular mechanism of GBM is conductive to advance therapeutic strategy and thus prolongs the survival.
LncRNAs (long non-coding RNAs) are over 200 nucleotides long, which are functional in epigenetic regulation, histone modi cation, transcription control and RNA metabolism [9,10].Current evidences have proven the regulatory effects of lncRNAs on cancer progression [11,12].L Salmena et al. [13] rst proposed ceRNA theory, in which lncRNAs exert a sponge effect on miRNAs, therefore upregulating mRNAs that are targeted to the certain miRNAs.Recently, the lncRNA-miRNA-mRNA axis has been widely explored in multiple types of cancers.Z Sun et al. [14] suggested that MALAT1 promotes EMT and angiogenesis in colorectal carcinoma by sponging miR-126-5p.H Wang et al. [15] proved that the XIST-miR-186-5p axis stimulates the proliferative and invasive potentials of NSCLC.J Zheng et al. [16] identi ed the role of CRNDE in aggravating glioma by regulating miR-384.
Recent studies have found that LBX2-AS1 is an oncogenic lncRNA that is highly expressed in gastric cancer, ovarian cancer and NSCLC [17][18][19].Besides, LBX2-AS1 is also upregulated in glioma samples, and correlated to the prognosis [20,21].This study rst detected the differential level of LBX2-AS1 in GBM tissues, and its oncogenic role and the underlying mechanism were further explored.

Collection of GBM samples
Thirty GBM samples were collected during surgery, including 20 primary GBM cases and 10 recurrent cases, and they were pathologically con rmed.In addition, 5 normal brain samples were collected during decompression surgery of traumatic brain injury at Neurosurgery Department, the First A liated Hospital of Nanjing Medical University, Nanjing, China.This study was reviewed and approved by the Clinical Research Ethic Committee of the First A liated Hospital of Nanjing Medical University.

Cell culture
Human glioma cell lines (U87, LN229, A172, T98G, U251), normal human astrocytes (NHA) and human umbilical vein endothelial cells (HUVECs) were provided by the American Type Culture Collection (ATCC).Primary GBM cell line (N3) was gifted from Beijing Tiantan Hospital.HEK293T cell line was provided by Cell Bank of the Chinese Academy of Sciences.Except for HUVECs cultured in EGM-2 (Lonza), the remaining were cultivated in DMEM containing 10% FBS.Cells were incubated at 37℃ in a humidi ed environment containing 5% CO 2 .
qRT-PCR Cells were lysed in TRIzol (Invitrogen, CA, USA) and the isolated RNA was reversely transcribed to cDNA using the PrimeScript RT (Takara, Nanjing, China).PARIS TM Nuclear/Cytosol Fractionation Kit was used for isolating nuclear and cytoplasmic components.After preparing a PCR system using the SYBR Green Premix Ex Taq (Takara, Nanjing, China), PCR was conducted with U6 or GAPDH as the internal reference.
Relative level was calculated using 2 -ΔΔCt method.Primer sequences are shown in Additional le 2: Table s4.
Fluorescence in situ hybridization (FISH) RNA-FISH was conducted as previously described [22].LBX2-AS1 probe was provided by RiboBio (Guangzhou, China).FISH-RNA signal in GBM cells and specimens were captured using a confocal microscope system (Zeiss LSM 700).

Cell proliferation assay
Proliferative potential of GBM cells was assessed by CCK-8, colony formation and EdU assay.In the rst experiment, cells were seeded in a 96-well plate and absorbance at 450 nm was measured using Cell Counting Kit-8 (Dojindo, Shanghai, China).In colony formation assay, cells were seeded in a 6-well plate and cultivated for 14 days.Colonies were washed in PBS twice, xed in 4% paraformaldehyde for 10 min and stained in 0.1% crystal violet for 30 min, which were captured under a microscope.EdU Apollo DNA in vitro kit (RiboBio, Guangzhou, China) was used for measuring DNA synthesis during the proliferative process of GBM cells.Cells seeded in a 6-well plate were incubated with 50 µM EdU for 2 h, followed by 30-min xation in 4% paraformaldehyde and staining with Apollo and Hoechst 33342.EdU-positive cells were captured using a uorescence microscope for counting.

Transwell assay
Transwell inserts (Corning, New York, USA) were pre-coated with 20 μg/μl Matrigel (BD Biosciences, New jersey, USA).2×10 4 cells suspended in serum-free medium and culture medium containing 10% FBS were respectively applied at the top and bottom of the prepared insert.After 24-h cell culture, cells invaded from the top to the bottom were xed in 4% paraformaldehyde for 10 min and dyed in 0.1% crystal violet for 30 min.Invasive cells in 3 random elds per sample were captured for counting.
3D tumor spheroid invasion assay 3D tumor spheroid invasion assay was conducted as previously described [24].2×10 5 cells in a 96-well plate were cultivated in complete medium containing spheroid formation matrix for 96 h, followed by adding Matrigel for embedding spheroids.Spheroids were captured at 0, 24, 48 and 72 h, respectively under a microscope, and the invasive area was evaluated using spheroids at 0 h as the reference.
Tube formation assay 1×10 4 HUVECs cultivated in conditioned medium of GBM cells were collected and seeded on a μ-Slide Angiogenesis coated with 10 μl Matrigel (BD Bioscience, New Jersey, NJ, USA) at 37℃ for 30 min.After 3-h cell culture, tube formation was captured under an optical microscope.

Dual-luciferase reporter assay
Promoter-containing vector for LBX2-AS1, Sp1, TEAD2 and KLF5 (Genechem, Shanghai, China) were cotransfected into HEK293T cells, respectively.P1-wt (wild-type) and P1-mut (mutant) sequences were synthesized and cloned into pGL3-basic luciferase vectors (Promega, Madison, USA), which were cotransfected into HEK293T cells with Sp1 plasmid.In addition, luciferase vectors containing mutant or wild-type sequences in which miR-491-5p bound to promoter region of LBX2-AS1 or LIF were cotransfected into N3 or U87 cells with miR-491-5p mimic or negative control.Relative luciferase activity was measured using the Promega Dual-luciferase Reporter System, and normalized to that of Renilla luciferase activity.

Chromatin immunoprecipitation (ChIP)
EZ-ChIP Kit (Millipore, Billerica, MA, USA) was used for ChIP assay.Brie y, cells were cross-linked in 1% formaldehyde for 10 min and terminated by glycine.Cells were lysed to chromatin fragments by sonication.DNA/protein complex was incubated with 3 μg anti-Sp1 (Cell Signaling Technology, 9389).Anti-IgG (Millipore, 12-371) was used as the negative control.The special primers for P1 site, P2 site and P3 site are listed in Additional le 2: Table s5.

RNA immunoprecipitation (RIP)
Magna RIPTM RNA-Binding Protein Immunoprecipitation Kit (Millipore, Billerica, MA, USA) was used for RIP assay.Brie y, cell lysate was incubated with anti-Ago2(Abcam, ab32381), and anti-IgG (negatively control).A protein-RNA complex was captured, followed by removal of the protein.The magnetic beads were repeatedly washed with RIP washing buffer, and the immunoprecipitated RNA was quanti ed by performing qRT-PCR.

Immunohistochemistry (IHC)
After sacri ce, GBM tissues collected from nude mice were xed in 4% paraformaldehyde, and para n embedded for preparation of tissue sections.Sections were incubated with anti-Ki-67, anti-LIF and anti-p-STAT3 for IHC.
Orthotopic GBM xenograft model 6-week-old male BALB/c nude mice were provided by Animal Core Facility of Nanjing Medical University.
Twelve mice were used in the subcutaneous xenograft GBM model, with 6 mice in each group.They were respectively subcutaneously administrated with 1×10 7 U87 cells transfected with sh-NC or sh-LBX2-AS1#1.The growth of tumor was measured once a week and the volume was calculated: Tumor volume (mm 3 ) = 0.5 × the longest diameter (mm) × the shortest diameter 2 (mm 2 ).In the intracranial xenograft GBM model, 12 mice were randomly assigned to two groups, with 6 in each.They were respectively intracranially administrated with 2.5×10 5 U87 cells transfected with sh-NC or sh-LBX2-AS1#1 using a stereotaxis instrument.Tumors in mice was examined using optical imaging (IVIS spectrum, PerkinElmer, USA).

Statistical analysis
Statistical analyses were conducted using GraphPad software version 8.0 (GraphPad software, San Diego, CA, USA) or SPSS Statistics 23.0 (SPSS, Chicago, IL, USA).Differences between groups were compared by Student's t test or one-way ANOVA.Pearson's correlation test was performed for evaluating the correlation between two indexes.Kaplan-Meier survival analysis was performed, and the difference was compared by log-rank test.All experiments were repeated in triplicate, and data were expressed as mean ± standard error of the mean (SEM).A signi cant difference was considered at p<0.05 (*p<0.05,**p<0.01,***p<0.001).

Upregulation of LBX2-AS1 in glioblastoma
To identify differentially expressed lncRNAs between glioblastoma and normal samples, we analyzed RNA-seq data from TCGA, CGGA and GEO (GSE151352) using the limma R package (FDR < 0.05, |Log 2 FC| >1).Upregulated and downregulated lncRNAs in GBM samples were illustrated in heatmap and volcano plots (Fig. 1a, Fig. s1a).By selecting the intersection, LBX2-AS1, CRNDE, H19 and MIR210HG were predicted upregulated in all the three databases (Fig. 1b).We subsequently analyzed their expression levels in 693 glioma samples from CGGA.Only LBX2-AS1 was signi cantly different between primary and recurrent cases both in low-grade glioma (LGG) and GBM (Fig. 1c).In particular, LBX2-AS1 level was higher in GBM samples than that of low-grade glioma (Fig. 1d).By assessing clinical features of glioma samples in TCGA and CGGA, it is indicated that LBX2-AS1 was highly expressed in mesenchymal (MES) glioma samples (Fig. 1e-f, Fig. s1b-c, Table s6-7).It is generally considered that the transition from proneural (PN) to MES subtype is a hallmark for the malignant development and recurrence of glioma [25][26][27].In addition, Kaplan-Meier survival analysis obtained the conclusion that LBX2-AS1 was unfavorable to the overall survival of LGG and GBM (Fig. 1g).Collectively, LBX2-AS1 was upregulated in GBM samples, and of signi cance in the malignant progression of glioma.
We synthesized three LBX2-AS1 shRNAs and tested their transfection e cacy in N3 and U87 cells, and nally, sh-LBX2-AS1#1 was selected for its excellent performance (Fig. 2b).Moreover, transfection of pcDNA-LBX2-AS1 markedly upregulated LBX2-AS1 in N3, U87 and U251 cells (Fig. 2c).Similarly, LBX2-AS1 level was higher in GBM samples collected in our center than that of normal brain samples, and notably, it was remarkably higher in recurrent GBM samples compared with that of primary ones (Fig. 2d).
RNA-FISH for quanti cation of LBX2-AS1 in GBM samples, as expected, was consistent with PCR results (Fig. 2e).To assess the regulatory effect of LBX2-AS1 on malignant development of GBM, GBM samples from TCGA and CGGA were assigned to low group and high group, followed by determination of differentially expressed mRNAs and they were depicted in heatmap and volcano plots (Fig. 2f-g).Since LBX2-AS1 was an oncogenic lncRNA, upregulated mRNAs in GBM samples were more concerned.
According to the results yielded from gene set enrichment analysis (GSEA), high-level LBX2-AS1 was identi ed correlated to EMT, angiogenesis and extracellular matrix disassembly in GBM samples (Fig. 2h).In addition, the Verhaak subtype signatures identi ed that the upregulation of LBX2-AS1 was positively correlated to MES subtype, but negatively correlated to PN subtype of glioma (Fig. s1d).

Knockdown of LBX2-AS1 suppresses proliferative potential, angiogenesis and EMT of GBM
Colony formation assay revealed that knockdown of LBX2-AS1 in N3 and U87 cells markedly reduced colony numbers, indicating the suppressed proliferative potential (Fig. 3a).Cell viability of GBM was consistently reduced by transfection of sh-LBX2-AS1 (Fig. 3b).3D tumor spheroid invasion assay obtained the conclusion that knockdown of LBX2-AS1 inhibited invasive ability of GBM cell spheroids (Fig. 3c), which was further validated in transwell assay (Fig. 3d).Compared with negative control, the tube formation ability of HUVECs cultivated in conditioned medium of N3 and U87 cells transfected with either sh-LBX2-AS1#1 or sh-LBX2-AS1#2 was attenuated (Fig. 3e).To assess the in vivo regulatory effects of LBX2-AS1 on tumorigenesis of GBM cells, we established xenograft model in nude mice administrated with U87 cells intervened by LBX2-AS1.Compared with control group, knockdown of LBX2-AS1 obviously slowed down the growth rate of tumor in nude mice (Fig. 3f-g).We further detected EMT markers in GBM cells regulated by LBX2-AS1.Protein levels of N-cadherin, Vimentin were downregulated, while E-cadherin was upregulated in N3 and U87 cells with LBX2-AS1 knockdown (Fig. 3h).Overexpression of LBX2-AS1 obtained opposite expression changes of them (Fig. 3i).Taken together, knockdown of LBX2-AS1 suppressed cell proliferation, angiogenesis and EMT in GBM cells.

Sp1 upregulates LBX2-AS1 in GBM samples
We next explored why LBX2-AS1 was abnormally upregulated in GBM samples.The top 10 scored transcription factors that could bind to the promoter region of LBX2-AS1 were screened out using JASPAR (http://jaspar.genereg.net/)(Fig. 4a).Relative levels of them in GBM samples, and their correlation to LBX2-AS1 were assessed by RNA-seq data from TCGA and CGGA (Fig. 4b, Fig. s2a).Finally, three transcription factors intersected from the two databases, including Sp1, TEAD2 and KLF5 were selected for further investigations (Fig. 4c).Promoter-containing vector for LBX2-AS1 and transcription factors were co-transfected into HEK293T cells with negative control, followed by measurement of relative luciferase activity, and that of Sp1 remained the highest (Fig. 4d).A positive correlation was detected between relative levels of Sp1 and LBX2-AS1 in 30 clinical specimens of GBM, which was consistent with the correlation analysis in TCGA and CGGA (Fig. 4e, Fig. s2b).Knockdown of Sp1 markedly downregulated LBX2-AS1, while overexpression of Sp1 upregulated it in N3 and U87 cells (Fig. 4f-g, Fig. s2c).Through analyzing ChIP-seq data of transcription factors obtained from ENCODE database and putative transcription factors obtained from JASPAR in UCSC (http://genome.ucsc.edu/),Sp1 was found enriched in the promoter region of LBX2-AS1 (Fig. 4h).The top three scored potential sites, where Sp1 bound to LBX2-AS1 were named as P1, P2 and P3 (Fig. 4i).Later, ChIP assay con rmed that Sp1 was enriched in P1 site in HEK293T cells (Fig. 4j).After mutating P1 site in the promoter region of LBX2-AS1, overexpression of Sp1 did not signi cantly affect relative luciferase activity, proving the direct interaction between Sp1 and LBX2-AS1 at P1 site (Fig. 4k-l).Therefore, we have con rmed that Sp1 induced upregulation of LBX2-AS1 in GBM samples.

LIF advances malignant phenotypes of GBM cellsvia the LIF-STAT3 axis
According to the Hallmark and GO gene sets yielded from GSEA, high-level LIF was identi ed correlated to EMT, angiogenesis and extracellular matrix disassembly in GBM samples, which was consistent with those of LBX2-AS1 (Fig. s4a).Transfection e cacy of si-LIF and pcDNA-LIF in N3 and U87 cells was examined by both qRT-PCR and Western blot (Fig. s4b-c).A series of in vitro experiments were carried out to assess the effect of LIF as the component in the ceRNA network on triggering malignant phenotypes of GBM.Data of CCK-8 and EdU assay both concluded that overexpression of LBX2-AS1 could abolish the inhibitory effect of si-LIF on proliferative potential of GBM cells (Fig. 7a-c).Later, the suppressed invasive ability of N3 and U87 cells transfected with si-LIF was abolished by overexpression of LBX2-AS1 (Fig. 7df).Knockdown of LIF markedly inhibited angiogenesis of GBM, but it was reversed by overexpression of LBX2-AS1 (Fig. 7g).Serving as a cytokine of the IL-6 family, LIF activates the JAK/STAT3, MAPK and PI3K signaling pathways by binding to the speci c receptor LIFR [30].The activation of the LIF/STAT3 signaling results in nuclear translocation of phosphorylated STAT3, thus initialing the transcription of downstream genes that are responsible for mediating EMT, angiogenesis and other biological activities.
In vivo knockdown of LBX2-AS1 slows down the growth of glioma in nude mice.
To assess the in vivo function of LBX2-AS1 in GBM, 2.5×10 5 luciferase-labeled U87 cells transfected with sh-LBX2-AS1#1 or sh-ctrl were intracranially administrated in nude mice.The growth of tumor was evaluated using optical imaging.Compared with control group, the volume of xenograft tumor tissues collected from LBX2-AS1 knocking-down group was remarkably smaller, and the survival rate was much higher (Fig. 8a-b).IHC results showed lower positive expressions of Ki-67, LIF and p-STAT3 in tumor sections from LBX2-AS1 knocking-down group than those of controls (Fig. 8c).

Discussion
Cancer-associated functions of lncRNAs have been highlighted, and they are potential biomarkers or therapeutic targets [11,31].Currently, the potential function of lncRNAs in GBM has been widely investigated [32,33].Through bioinformatic analyses using online databases, we have found that LBX2-AS1 was upregulated in GBM samples, and positively correlated to tumor staging of glioma.A high level of LBX2-AS1 predicted poor prognosis of glioma.Notably, LBX2-AS1 level was much higher in recurrent GBM cases than that of primary ones, indicating a potential relationship between LBX2-AS1 and GBM recurrence.The following in vitro experiments have demonstrated that LBX2-AS1 promoted proliferative, invasive and angiogenic capacities of GBM, serving as an oncogenic lncRNA.We believed that LBX2-AS1 was a promising biomarker that could be applied to the diagnosis and treatment of GBM.Some lncRNAs are dysregulated in human cancers, however, why they are dysregulated have not been fully elucidated.The transcription of lncRNAs is mediated by transcription factors and epigenic regulatory factors [34][35][36].Using JASPAR and ENCODE, multiple Sp1 binding sites were identi ed in the promoter region of LBX2-AS1.Furthermore, Sp1 was con rmed to activate the transcription of LBX2-AS1 through ChIP and dual luciferase reporter assay.Subcellular distribution determines the role of lncRNAs in regulating malignant phenotypes of cancer cells [37,38].LBX2-AS1 was both expressed in cell nuclei and cytoplasm, which was more pronounced in the latter.Cytoplasmic lncRNAs have been widely recognized for their ceRNA function.By competitively sponging miRNAs, lncRNAs prevent the Ago-dependent degradation of target mRNAs that bind to miRNAs [28,39].As RIP assay shown, LBX2-AS1 could interact with Ago2, suggesting that LBX2-AS1 was able to exert the sponge effect.Subsequently, miR-491-5p was proven to be a potential target of LBX2-AS1.It is reported that miR-491-5p can alleviate the malignant progression of glioma [40].Our ndings showed a negative correlation between relative levels of LBX2-AS1 and miR-491-5p.Importantly, the suppressed proliferative, invasive and angiogenic potentials of GBM cells by knockdown of LBX2-AS1 could be reversed by silence of miR-491-5p.Hence, we have proven that LBX2-AS1 exerted its oncogenic role in GBM by sponging miR-491-5p as a ceRNA.
According to the prediction using starBase, LIF was identi ed as a potential target of miR-491-5p, which was further con rmed by experiments.LBX2-AS1 could positively regulate LIF level in GBM cells, while LIF level was negatively regulated by miR-491-5p.Previous evidences have shown the upregulation of LIF in solid tumor samples, which mediates the growth, invasiveness and metastasis of tumor cells [41].Through activating the STAT3 signaling, LIF boosts the proliferative and invasive capacities of choriocarcinoma cells [42].The self-renewal of glioma-initiating cells is achieved by the LIF/STAT3 signaling [43].Shi et al. [44] revealed the biological signi cance of LIF in triggering the development and chemotherapy resistance of pancreatic ductal adenocarcinoma.This study rst reported that LBX2-AS1 regulated the LIF/STAT3 signaling as a ceRNA in GBM.STAT3 is a transcription factor that is well recognized for mediating downstream gene transcription, such as those involved in the regulation of EMT and angiogenesis.Jin W et al. [45] proposed that STAT3 stimulates EMT in tumor cells by upregulating Snail and Slug.VEGF and MMP9 are key mediators of tumor angiogenesis, which are regulated by STAT3 [46,47].Our results showed that knockdown of LIF inhibited the activation of the LIF/STAT3 signaling, manifested as downregulated Snail, Slug, MMP9 and VEGF.Meanwhile, angiogenesis and EMT of GBM were suppressed as well, which could be reversed by overexpression of LBX2-AS1.
Two limitations in the present study should be considered.Firstly, we proved that Sp1 could bind to the promoter region of LBX2-AS1 and thus regulated its transcription.The speci c mechanism, however, remains unclear.Secondly, LBX2-AS1 was highly expressed in recurrent GBM samples.The potential interaction between LBX2-AS1 and drug resistance of GBM needs further explorations.

Conclusions
LBX2-AS1 is upregulated in GBM and correlated to the poor prognosis.It exerts the sponge effect on miR-491-5p, thus upregulating LIF and activating the LIF/STAT3 signaling.The LIF/STAT3 signaling is responsible for promoting proliferation, EMT and angiogenesis of GBM.In addition, Sp1 is identi ed to induce the expression of LBX2-AS1, although the underlying mechanism needs further investigation.Our ndings suggested that LBX2-AS1 may be a novel prognostic marker and therapeutic target of GBM.

Figures
Figures