mRNA expression datasets and the corresponding clinical information were downloaded from the following websites: Chinese Glioma Genome Atlas (CGGA) (http://www.cgga.org.cn), the Cancer Genome Atlas (TCGA) (http://cancergenome.nih.gov), and the REpository for Molecular BRAin Neoplasia DaTa (REMBRANDT) (http://caintegrator.nci.nih.gov/rembrandt/).

The human GBM cell lines U87, U87vIII and U251 were used for experiments. U87 and U251 GBM cells were purchased from the ATCC (Manassas, VA, USA). U87 cells were cultured in complete MEM medium, while U251 cells were cultured in complete EMEM medium containing 10% FBS, and incubated at 37°C, 5% CO₂. U87vlll cells were stably transfected with a truncated mutant EGFR gene, which can be consistently activated without EGF stimulation. Puromycin (100 ng/ml) was added into U87vIII culture medium to maintain the stability of EGFRvIII.

To determine mRNA expression, after treatment, the glioma cells were lysed in TRIzol reagent (Sigma-Aldrich, St. Louis, MO, USA). The lysate was well mixed with chloroform and spun for 15 min at 13,000 × g at 4°C. The top aqueous phase, which contains RNA, was collected and mixed with propan-2-ol to precipitate the total RNA, which was used for real-time PCR analysis of mRNA. The cDNA was synthesized from 1 µg of total RNA, using a reverse transcription kit purchased from Promega (Madison, WI, USA), following the manufacturer's protocol. Real-time PCR was performed using a SYBR Green Master Mix from Life Technologies (Carlsbad, CA, USA). Amplification was performed by the DNA Engine Opticon 2 Two-Color Real-time PCR detection system (Bio-Rad Laboratories, Hercules, CA, USA). Relative gene expression was analyzed by 2ΔΔ-Cq method (22). Gene specific primers: HOTAIR-F: ATAGGCAAATGTCAGAGGGTT, HOTAIR-R: TCTTAAATTGGGCTGGGTC; GAPDH-F: GGTGGTCTCCTCTGACTTCAACA, GAPDH-R: GTTGCTGTAGCCAAATTCGTTGT; CCNA2-F: CTCTACACAGTCACGGGACAAAG, CCNA2-R: CTGTGGTGCTTTGAGGTAGGTC; HMMR-F: GGCTGGGAAAAATGCAGAGGATG, HMMR-R: CCTTTAGTGCTGACTTGGTCTGC; FoxM1-F: CCTTCTGGACCATTCACCCC, FoxM1-R: TCACCGGGAACTGGATAGGT; NUSAP1-F: CTGACCAAGACTCCAGCCAGAA, NUSAP1-R: GAGTCTGCGTTGCCTCAGTTGT; ASPM-F: GAGACCTTGGTGGAATACCTGC, ASPM-R: ACGAAGATCCAAAAGCCTTGCAC; CDC6-F: TCCACCAAAGCAAGGCAAGA, CDC6-R: CGATCTGGGACAGCTGTGTT; FANCI-F: GCAAGCTGATGTTCGACTCATGC, FANCI-R: AGGCAGCAGATCAGGTTTTGGC; NCAPG-F: GACGAACAGGAGGTGTCAGACT, NCAPG-R: TGCTGCGGTTTTGGCTCGTCTT; DLGAP5-F: CTCGATCAGCTACTCAAGCAGC, DLGAP5-R: CAGGTCTTCCTTTACTTGGCACC; CHEK1-F: ATCAACTCATGGCAGGGGTG, CHEK1-R: TCCAGCGAGCATTGCAGTAA; CEP55-F: TCG ACCGTCAACATGTGCAGCA, CEP55-R: GGCTCTGTGATGGCAAACTCATG; KIF4A-F: TGCGTGGTCAAGTTT CGGAGTC, KIF4A-R: GCTGTAGGTCAGCAATCTGAGC; HJURP-F: TGAGAATTTGGGGTGGAAGACT, HJURP-R: AGCGGAGTCACACGTACATC; PLK4-F: GACACCTCAGACTGAAACCGTAC, PLK4-R: GTCCTTCTGCAAATCTGGATGGC; CCNB2-F: AGTTCCAGTTCAACCCACCAA, CCNB2-R: TTGCAGAGCAAGGCATCAGA; CENPE-F: GGAGAAAGATGACCTACAGAGGC, CENPE-R: AGTTCCTCTTCAGTTTCCAGGTG; NCAPH-F: GACGAACAGGAGGTGTCAGACT, NCAPH-R: TGCTGCGGTTTTGGCTCGTCTT; AURKB-F: CATCCCAACATCCTGCGTCT, AURKB-R: AGCTCTCCCTTGAGCCCTAA.

Protein lysates were prepared as follows: after treatment, cells were washed twice with cold PBS, scraped and lysed in ice-cold RIPA buffer (Solarbio, Beijing, China). Lysate was sonicated for 20 cycles, and microcentrifuged for 15 min at 4°C. The supernatant was transferred to a new test tube and stored at 20°C. The protein samples were resolved by SDS-PAGE and transferred onto PVDF membranes (Milipore, Darmstadt, Germany). The membranes were then incubated with the following antibodies: anti-FoxM1 (Cell Signaling Technology, Danvers, MA, USA), anti-CEP55 (Abcam, Cambridge, UK) and GAPDH (Proteintech, Wuhan, China). Antibody-labeled protein bands on the PVDF membranes were detected using a G:BOX F3 (Syngene, Cambridge, UK).

U87 cells were seeded onto the coverslips and transfected with negative control or Lenti-HOTAIR virus (GeneChem, Shanghai, China) for 48 h. Then, they were fixed in 4% paraformaldehyde for 30 min. The cells or sections were permeabilized with 0.1% Triton-X 100 for 10 min, followed by blocking with 3% BSA for 1 h in RT. Immunofluorescence staining was conducted with antibodies against CEP55 (Abcam, 1:100) and CENPE (Abcam, 1:100). The cells were washed with PBS and incubated with Alexa Fluor 633 or Alexa Fluor 594 (Life Technologies) secondary antibodies. Nuclei were stained using DAPI and the cells were visualized using FV-1200 laser scanning confocal microscope.

U87vIII cells were transfected with either a negative control or Lenti-si-HOTAIR virus. An indicated number of cells in suspension were injected stereotactically into the brain of 4-week-old BALB/c-nu mice. The mice were sacrificed at day 14.

The xenograft samples were collected at day 14 after tumor implantation, and were subjected to Hematoxylin and eosin (H&E) staining. Sections (5 µm) were cut, dehydrated, deparaffinized, and rehydrated. H&E staining was performed according to the standard protocols. All images were captured via microscopy (Olympus, Tokyo, Japan). The protocol for the animal study was approved by the Animal Ethics Committee of Tianjin Medical University.

Statistical analysis was performed using the SPSS Graduate Pack, version 11.0, statistical software (SPSS). Data are presented as means ± SEM of three independent experiments or means ± SD performed in triplicate. One-way ANOVA was used for comparison among the different groups. A P-value of 0.05 was considered to indicate a statistically significant difference.

Although variety of lncRNAs have been described in recent years, the functions of lncRNA are still not completely characterized. To profile lncRNAs in glioma, cluster analysis was employed using the CGGA database, which includes 220 glioma samples and 5 normal brain samples. There are 90 lncRNAs that have different patterns of expression between low-grade and high-grade gliomas. There are 31 lncRNAs downregulated in high-grade glioma, while 59 lncRNAs were upregulated (FDR <0.05, fold >1.5) (Fig. 1). It is possible that the main alteration of lncRNA expression occurs during malignant progression from low-grade glioma to GBM, indicating an important role of lncRNA in glioma progression. HOTAIR is among the lncRNAs in the high expressin module of the heat map, which is consistent with a previous study (15). HOTAIR is a cell cycle-associated oncogene in glioma (11), however, further research is still needed to demonstrate its role in gliomagenesis.

To profile the function of HOTAIR, genes positively correlated with HOTAIR were selected by Pearson correlation coefficients (R>0.3, P<0.05) (Fig. 2A). In total, 244 HOTAIR correlated genes formed a complex network, in which they correlated with each other according to literature, databases or experiments. Genes with P-values <0.01 were extracted from the HOTAIR positive gene network, as they are more significantly correlated with HOTAIR (Fig. 2B). The correlation degrees of the 18 genes are greater than 25, and all of them interact with each other. Overall, 215 of the 244 genes in network are directly correlated with these 18 genes, implying a key role of this 18-gene network. The cluster analysis of these genes according to HOTAIR expression in the CGGA, REMBRANDT and TCGA databases further confirmed the positive correlation expression pattern of these 18 genes with HOTAIR (Fig. 2C–E). Bioinformatic analysis revealed that the upregulation of 17 of them (except CCNB2) corresponds with clinical stage glioma (Table I). Kaplan-Meier survival analysis further indicated a negative correlation between the expression level of 17 genes and the survival rate in high-grade glioma. Fig. 2F–M depicts four genes with the most significant P-values from the Kaplan-Meier survival analysis. Of these 18 genes, 15 genes play roles in cell cycle regulation, including CCNA2, FoxM1, CEP55, CENPE. FANCI and CDC6 are genes regulating DNA replication and repair, which may influence the cell cycle indirectly. Thus, we identified these mRNAs as a cell cycle-related network that positively correlated with HOTAIR.

To further test if this mRNA network could be upregulated by HOTAIR in vitro, we transfected a tetracycline-inducible HOTAIR expression lenti virus. Real-time PCR analysis indicated that genes in this network were upregulated when HOTAIR expression was induced by doxycycline in U87 (Fig. 3A) and U251 (Fig. 3B) cells. EGFRvIII is an aggressive EGFR mutation type in glioma. We knocked down HOTAIR in U87vIII cells by Lenti-si-HOTAIR and discovered that this network was downregulated by inhibiting HOTAIR (Fig. 3C). FoxM1 is a key transcription factor that regulates the expression of cell cycle genes (23,24). CEP55 and CENPE, which are important regulators of mitosis, impact survival rate very significantly in high-grade glioma (HGG) (24,25). Thus, we representatively chose FoxM1, CEP55 and CENPE from the network to further examine the influence of HOTAIR on these genes and the proteins they code. Western blot analysis indicated that HOTAIR could upregulate FoxM1 and CEP55 in U87 cells, while Lenti-si-HOTAIR-treatment inhibited expression of these proteins in U87vIII cells (Fig. 3D). Immunofluorescence staining further confirmed the upregulation of CEP55 (Fig. 3E) and CENPE (Fig. 3F) in U87 cells. Thus, we demonstrated that this cell cycle-related mRNA network is positively regulated by HOTAIR.

EGFR is a receptor tyrosine kinase that is frequently amplified and mutated in several cancers. EGFRvIII is one such mutation commonly found in GBM patients, and it contributes to the malignant progression of the disease (26). Knocking down HOTAIR significantly inhibited the cell cycle-related mRNA network in U87vIII cells. Thus, we targeted HOTAIR in U87vIII intracranial GBM mouse models. U87vIII cells (2×10⁵) could form orthotopic tumors in mice in 14 days. However, after lenti-si-HOTAIR treatment, there was no macroscopic tumor formation even after the injection of 8×10⁵ cells (Fig. 4A). These results indicated that knocking down HOTAIR could control U87vIII tumor formation within 2 weeks. Our bioinformatic analysis of the CGGA database further indicated that HOTAIR expression is upregulated in high-grade glioma (Fig. 4B), and high levels of HOTAIR are correlated with poor outcome in glioma (Fig. 4C) and HGG (Fig. 4D). These results indicated that HOTAIR could be a potential therapeutic target in glioma, especially for GBM.