All 60 tissue samples were collected from Harbin Medical University (Heilongjiang, China). The present study received ethical approval from Harbin Medical University. The clinical characters of these samples were listed in Table I. Written informed consent approving this study was obtained from each patient. Human glioblastoma cell lines U-87MG (www.atcc.org/Products/All/HTB-14.aspx) and U251 were from the American Tissue Culture Collection (ATCC, Manassas, VA, USA). The U-87MG cell line is known to be cross-contaminated with another cell line of unknown origin, but it is likely to be another glioblastoma cell line (20). U-87MG cells were grown in Minimum Essential Medium (HyClone; GE Healthcare Life Sciences, Logan, UT, USA) with 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA). U251 cells were maintained in Dulbecco's High Glucose Modified Eagle Medium (HyClone; GE Healthcare Life Sciences) with 10% FBS. Normal human astrocytes (NHA) obtained from Lonza (www.lonza.com/products-services/bio-research/primary-cells/human-cells-and-media/neural-cells-and-media/nha-normal-human-astrocytes.aspx) were cultured in the provided astrocyte growth media and 5% FBS. Cells were incubated in a humidified atmosphere with 5% CO₂ at 37°C.

NC mimic, miR-500a-5p mimic and its corresponding miR-500a-5p inhibitor were purchased from Guangzhou RiboBio Co., Ltd., (Guangzhou, China). Cell transfection was performed with Lipofectamine 2000 Reagent (Invitrogen; Thermo Fisher Scientific, Inc.) according to manufacturer's protocol. After incubating, cells were harvested then experiments were performed. Loss of CHD5 expression was achieved using small interfering RNA (siRNA) of CHD5. To establish a stably CHD5-silenced cell line, a GV113 plasmid containing CHD5 lentivirus short hairpin RNAs (TGGTTAAGGGCAGTGATAG) were separately transduced into U-87MG cells.

Cell proliferation was detected by Cell Counting Kit (7 sea biotech, Shanghai, China). Cells were grown in 96-well plate with 1×10⁴ per well and incubated in 37°C with 5% CO₂ until cell confluent rate reached 70%. After transfected with plasmid for 48 h, cells were still incubated for 24, 48 and 72 h. 10 µl CCK8 solution was seed into each well. The absorbance at 450 nm was measured with SUNRISE Microplate Reader (Tecan Group, Ltd., Mannedorf, Switzerland).

Cell apoptosis was measured by flow cytometry following the instructions of the Annexin V-FITC/PI apoptosis detection kit (Nanjing KeyGen Biotech Co., Ltd., Nanjing, China). Cultured cells were harvested and washed twice in PBS, re-suspended in binding buffer, and then incubated with Annexin V-FITC and propidium iodide (PI) for 15 min at room temperature. Afterwards, flow cytometry was performed to determine rate of apoptosis on a FACSAria flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA).

Twenty-four-well Transwell chambers with 8-µm pore size polycarbonate (Corning Incorporated, Corning, NY, USA) were used for cell migration and invasion assays. For invasion assays, the top side of the membrane was coated with Matrigel (BD Biosciences), and then 1×10⁵ cells (in each well) in serum-free DMEM or RPMI 1640 medium were seeded on the chambers. DMEM or RPMI 1640 containing 10% FBS was added to the wells under the chamber. For migration analysis, 5×10⁴ cells (in each well) in serum-free DMEM or RPMI 1640 medium were seeded on the chambers without Matrigel. After 24 h of incubation, cotton swabs were used to remove the cells inside the upper chamber, while the cells on the other side of the membrane surface were fixed and stained with 0.5% crystal violet solution. Five random fields were counted in each well.

The potential targets of miR-500a-5p were predicted using the TargetScan (www.targetscan.org), microRNA (www.microrna.org/) and miRDB (www.mirdb.org). The common genes of these algorithms were selected for further analysis.

The total RNA of the tissue samples and cells was isolated using TRIzol Reagent (Invitrogen; Thermo Fisher Scientific, Inc.), in accordance with manufacturer's instructions. Then, 1 µg of total RNA was reverse-transcribed in a volume of 20 µl using random and oligo dT primers under standard conditions, in accordance with the instructions of the PrimeScript RT kit (Takara Biotechnology Co., Ltd., Dalian, China). For RT-qPCR assays, we used SYBR Premix Ex Taq (Takara Biotechnology Co., Ltd.) to determine the expression level of CHD5, in accordance with manufacturer's instructions. The thermocycling conditions were as follows: 95°C for 30 sec, followed by 40 cycles at 95°C for 15 sec and 60°C for 32 sec and dissociation at 95°C for 60 sec, 55°C for 30 sec and 95°C for 30 sec. The expression data of CHD5 were normalized to the expression of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The primers used were as follows: Forward primer, 5′-CTGACTTCAACAGCGACACC-3′ and reverse primer, 5′-TCTGACTTCAACAGCGACACC-3′. The relative fold change was calculated using the 2^−ΔΔCq method (21).

The protocol was described previously (22). Four-week-old athymic BALB/c nude mice were maintained under specific pathogen-free conditions. The mice were manipulated in accordance with the protocols approved by Harbin Medical University. The U-87MG cells were harvested and washed with PBS. Then, 1×10⁷ cells were subcutaneously injected into the ventral side of each mouse for tumor formation assays. Six mice were used for each group. The tumor volumes were examined every 7 days and calculated. The protocol was approved by the Animal Care Ethics Committee of Harbin Medical University.

In this dataset, 16 fresh-frozen glioblastoma multiforme samples, 7 healthy brain tissues, a NHA cell line and human fetal astrocyte cell line were analyzed by using miRNA and whole transcriptome microarray chips.

U-87MG cells were seeded into a 24-well plate. Cells were co-transfected with wild-type, mutated CHD5 reporter plasmid or pMIR vector, and miR-500a-5p mimics or miR-500a-5p inhibitor. Luciferase assays were conducted 24 h after transfection using the Dual Luciferase Reporter Assay System (Promega Corporation, Madison, WI, USA).

Statistical analysis was performed using SPSS v18.0 software (SPSS, Inc., Chicago, IL, USA). Each experiment was repeated at least three times. All data are expressed as the mean ± standard deviation. The Kaplan-Meier method was used to calculate the survival curve, and log-rank test to determine statistical significance. Student's t-test and one-way analysis of variance followed by Tukey's post hoc test were used to analyze 2 or multiple groups, respectively, for statistical significance. A paired Student's t-test was used to compare differences between human glioblastoma and normal adjacent tissue samples. Diagnostic potential of miR-500a-5p expression was assessed by receiver operating characteristic (ROC) analysis. The respective area under the curve (AUC) was analyzed by the Hanley and McNeil method. For analysis of correlation between miR-500a-5p levels and clinical features, chi-square tests were used. Pearson correlation coefficient analysis was used to determine the correlations. P<0.05 was considered to indicate a statistically significant difference.

To explore the function miR-500a-5p, we firstly examined its expression in glioblastoma tissues by RT-qPCR. We found that miR-500a-5p was significantly upregulated in glioblastoma tissues compared with adjacent normal tissues (Fig. 1A). Similarly, the high expression of miR-500a-5p was verified in glioblastoma cell lines (Fig. 1B). Then we divided these samples into two groups according to miR-500a-5p expression levels (mean value was the cut-off). Kaplan-Meier analysis was used to evaluate survival. The results showed that patients with higher miR-500a-5p expression possessed poorer survival (Fig. 1C). Furthermore, ROC curves were performed to evaluate the sensitivity and specificity of miR-500a-5p expression in predicting glioblastoma tissues from normal tissues. Notably, miR-500a-5p displayed predictive, with an AUC of 0.753 (P = 0.046; Fig. 1D). These results implied that miR-500a-5p might provide imperative clinical significance in glioblastoma diagnosis.

To determine the functions of miR-500a-5p in glioblastoma, we overexpressed miR-500a-5p in U-87MG and U251 cells by transduction with NC mimic or miR-500a-5p mimic. At 48 h after transduction with miR-500a-5p, miR-500a-5p expression was significantly increased compared with NC mimic group (Fig. 2A). Then CCK8 assay was conducted to analyze the proliferation of U-87MG and U251 cells at 24, 48 and 72 h post transfection (Fig. 2B). Overexpression of miR-500a-5p observably promoted cell proliferation in both U-87MG MG and U251 cells. Similarly, colony formation assay showed that cells transfected with miR-500a-5p formed more clones (Fig. 2C). In consistence, we found that more U-87MG and U251 cells transfected with miR-500a-5p entered into S phase than control (Fig. 2D). To evaluate the effect of miR-500a-5p on cell migration and invasion, we performed transwell assays. We found that overexpression of miR-500a-5p promoted cell migration and invasion (Fig. 2E and F). In addition, we inhibited miR-500a-5p in U-87MG and U251 cells to evaluate the effect on cell apoptosis. Flow Cytometry by Annexin V/PI staining showed that apoptotic ratio was significantly higher in miR-500a-5p inhibitor-transfected cells than control (Fig. 2G). In collection, miR-500a-5p promoted cell proliferation, migration and invasion in glioblastoma, but inhibited cell apoptosis.

To further determine the effects of miR-500a-5p on glioblastoma cells in vivo, we examined tumor growth in nude mice. U-87MG cells transfected with NC inhibitor or miR-500a-5p inhibitor were subcutaneously inoculated into the armpits of the recipient nude mice. Every other 7 day, we measured the tumor volumes. We found that miR-500a-5p knockdown significantly inhibited tumor growth in vivo (Fig. 3A). At the endpoint of the experiments, the tumor weights were measured. The tumors in miR-500a-5p knockdown group were significantly lighter than that in control group (Fig. 3B).

To explore the molecular mechanism through which miR-500a-5p promoted glioblastoma progression, three computational algorithms including TargetScan, miRanda and PicTar were used in combination to search for potential targets of miR-500a-5p. Among the candidates, the chromatin remodeler and tumor suppressor CHD5 was predicted to be a target of miR-500a-5p. The predicted interaction between miR-500a-5p and 3′UTR of CHD5 was illustrated (Fig. 4A). In order to verify this prediction, we cloned 3′UTR-wt and 3′UTR-mut into pMIR-REPORT vector. As expected, dual luciferase assay demonstrated that miR-500a-5p overexpression remarkably inhibited the luciferase activity while miR-500a-5p knockdown increased the luciferase activity (Fig. 4B). In contrast, the effect of miR-500a-5p on luciferase activity observed in pMIR-3′-UTR-wt was absent in pMIR-3′-UTR-mut (Fig. 4B). Moreover, we found that overexpression of miR-500a-5p decreased the mRNA and protein levels of CHD5 in U-87MG and U251 cells while miR-500a-5p knockdown got the inverse result (Fig. 4C and D). In addition, by RT-qPCR we found that the expression of miR-500a-5p was inversely correlated with that of CHD5 in glioblastoma tissues (Fig. 4E). Furthermore, the dataset (GSE90598) in GEO database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE90598) showed that the expression of CHD5 was significantly downregulated in glioblastoma tissues compared with normal tissues (Fig. 4F). Similarly, RT-qPCR also showed that the mRNA levels of CHD5 were lower in glioblastoma samples than normal tissues (Fig. 4G). These results indicated that CHD5 mRNA 3′-UTR is a specific functional target of miR-500a-5p in glioblastoma cells.

To investigate whether the regulation of cell proliferation, migration and invasion of glioblastoma cells by miR-500a-5p is executed via a CHD5-dependent manner, we co-transfected U-87MG and U251 cells with miR-500a-5p inhibitor and CHD5 siRNA. Compared with cells transfected with miR-500a-5p inhibitor, the cells transfected with both miR-500a-5p inhibitor and CHD5 siRNA exhibited a lower expression on CHD5 protein level (Fig. 5A). CCK8 and colony formation assays showed that CHD5 knockdown reversed the inhibitory effects by miR-500a-5p inhibition on cell proliferation potential in U-87MG and U251 cells (Fig. 5B and C). In addition, transwell assays showed that CHD5 knockdown reversed the inhibitory effects by miR-500a-5p inhibition on cell migration and invasion potential in U-87MG and U251 cells (Fig. 5D and E). Furthermore, CHD5 knockdown decreased the cell apoptosis induced by miR-500a-5p inhibition (Fig. 5F). What's more, CHD5 knockdown also increased tumor weights to the control level in vivo (Fig. 5G). Besides, WB assay with formed tumor tissues showed that CHD5 knockdown reversed the inhibitory effects by miR-500a-5p inhibition on cell proliferation, migration and invasion in vivo (Fig. 5H). Accumulating studies showed that Wnt/β-catenin signaling is indispensable for cancer development. We then performed WB assays with formed tumor tissues and found that miR-500a-5p inhibition downregulated Wnt/β-catenin signaling while CHD5 knockdown upregulated it (Fig. 5I), which indicated that miR-500a-5p/CHD5 regulated glioblastoma development and progression through Wnt/β-catenin signaling at least in part.