We downloaded clinical characteristics and the data of SPAG5 mRNA expression profile chip expression data from The Cancer Genome Atlas (TCGA) database (http://www.cbioportal.org), including 667 glioma specimens and 10 normal specimens. The RNA-seq level 3 data of the expression profile of these samples were downloaded and sorted and directly used for the analysis of the mRNA expression of SPAG5. For pathological analysis, the RNA-seq level 3 data from the 667 glioma specimens, which was divided into low-grade gliomas (LGGs, n = 515, WHO II and WHO III grade gliomas) and GBMs (n = 152). We use the Affy and Limma packages in the R language to standardize and T test our data and then filter according to P value <0.05 and |FC| ≥2. According to the median of the SPAG5 mRNA expression, the 667 glioma specimens were further divided into low SPAG5 expression group (n = 334) and high SPAG5 expression group (n = 333). The association between the mRNA expression level of SPAG5 and the overall survival time of glioma patients was then analyzed by Kaplan–Meier curves.

Glioma cell lines (U87 and U251) were purchased from Shanghai Genechem Co., Ltd. (Shanghai, China). Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Corning) supplemented with 10% fetal bovine serum (FBS; Ausbian) and cultured in a 5% CO₂ incubator at 37°C. To knock down SPAG5 and overexpress CDH2, the plasmids specifically expressing SPAG5 shRNA and CDH2 mRNA were constructed using pAdTrack-CMV plasmid (Addgene, Cambridge, MA, USA) as the vector. The plasmids expressing the mRNA or shRNA that are not targeting any known human gene were used as the negative control. The cells were transfected with shSPAG5, CDH2 mRNA, or control mRNA or shRNA by Lipofectamine 2000 (Invitrogen, CA, USA) according to the instruction.

The total RNAs were extracted with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s protocol and reversed to cDNA (Invitrogen, CA, USA). qRT-PCR was performed, and the sequences of the PCR primers are as follows: SPAG5 F: 5′-TTGAGGCCCGTTTAGATACCA-3′ and R: 5′-GCTTTCCTTGGAGC-AATGTAGTT-3′; glyceraldehyde 3-phosphate dehydrogenase (GAPDH), F: 5’-TGACTTCAACAGCGACACCCA-3′ and R: 5’-CACCCTGTTGCTGTAGCCAAA-3′. The relative expression of SPAG5 was presented as 2^-ΔΔCt value.

Western blotting was carried out according to the literature description (17). The primary antibodies are as follows: rabbit polyclonal SPAG5 (1:200, Sigma-Aldrich, Germany), mouse monoclonal GAPDH (1:2,000, Santa-Cruz, CA, USA), rabbit polyclonal CDH2 (1:100, Cell Signaling Technology, MA, USA), rabbit IgG (1:2,000, Santa-Cruz, CA, USA). A horseradish peroxidase (HRP)-conjugated anti-rabbit or anti-mouse IgG antibody was used as the secondary antibody (1:2,000, Santa-Cruz, CA, USA).

The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay was used to measure cell proliferation rate. Glioma cells (2 × 10³ cells/well) were seeded onto 96-well plates. Subsequently, 20 μl MTT (5 mg/ml) solution was added to each well and incubated for 4 h at 37°C. After aspirating the medium, 100 μl dimethyl sulfoxide (DMSO) was added to solubilize the formazan crystals formed by viable cells. Optical density (OD) was measured at 490 nm. The observation duration lasted for 5 days.

Adherent glioma cells in the logarithmic phase were trypsinized and counted to measure viability. Then, viable cells (600 cells/well) were seeded onto each well of a six-well plate. Glioma cells were allowed to adhere and grow for 15 days. Media were replaced every 3 days. When colonies were formed, we removed media and added 1 ml 4% paraformaldehyde to each well to fix cells for 30 min and stained them with crystal violet solution (22). Finally, colonies were counted. Data gathered from three independent experiments were expressed as mean colony number ± SD.

Caspase 3/7 assay was used to assess apoptosis according to the manufacturer’s instructions (Caspase-Glo^® 3/7 Assay, Promega Corporation, Cat. No. G8092).

The 5 × 10⁴ glioma cells were inoculated in 96-well plate. When the cells grew to 90% confluence, we scratched the bottom of the dishes across each well using a scratch tester. Cells were rinsed 2–3 times with serum-free medium and cultivated in 0.5% FBS. The wound-healing process was observed for 24 h, and photos were taken at 8 and 24 h.

The cell migration and invasion assays were carried out by Transwell kit (Corning, US) following the manufacturer’s instructions. In brief, some chambers were inserted in a new 24-well plate, and 5 × 10³ cells in 100 μl medium without FBS were seeded on the upper chamber in Transwell apparatus. Then, 600 μl medium with 10% FBS was added in the lower chamber. After the cells were incubated for 16 h at 37°C, the medium in chambers were removed with absorbent paper, and cells on the chambers were wiped off with a cotton swab. The cells adhering to chambers were treated with 4% paraformaldehyde for 30 min to fix and stained with Giemsa solution and counted and visualized under a microscope in nine random fields (×200). The process of the invasion assay was similar to the cell migration experiment, except that the Transwell membrane was precoated with Matrigel basement membrane and the cells were cultured for 18 h at 37°C. The cell count method was the same as the cell migration assay.

Cells in the logarithmic growth phase were trypsinized and completely resuspended into cell suspension and counted. The cells (1,500 cells/well) were seeded onto each well of a 96-well plate and cultured at 37°C with 5% CO₂. The day after planking, the number of green fluorescent cells was counted under Celigo cytometry system (Nexcelom, Beijing, China) for 5 consecutive days. Lentivirus 2000 was used to infect cells to make the glioma cells express SPAG5 and green fluorescent proteins, so as to facilitate the automatic cell count. To further explore the effect of SPAG5 knockdown, CDH2 was overexpressed. The experiment was divided into the following: NC+NC group (parental glioma cells + vector), KD+NC group (parental glioma cells + knockdown-SPAG5 + vector), and KD+OE group (parental glioma cells + knockdown-SPAG5 + overexpression-CDH2).

All quantified data were obtained from at least three independent experiments and analyzed using SPSS 17.0 software. Data are shown as mean ± SD. The Kaplan–Meier method was used for survival analysis. The log-rank test was used to assess differences in survival. Spearman’s method was applied in analyzing the relationship between gene expression level and clinical variables. Comparison of gene expression between groups was conducted using Mann–Whitney U. The differences between groups were analyzed using two-tailed Student’s t-test and ANOVA. Differences were considered statistically significant when P < 0.05.

Analysis of the mRNA expression profiles of SPAG5 in TCGA revealed that the mRNA expression of SPAG5 was higher in glioma tissues (n = 667) than that in normal tissues (n = 10) (P < 0.05; Figure 1A). Meanwhile, the mRNA expression of SPAG5 was higher in GBMs (n = 152) than that in LGGs (n = 515) (P < 0.05; Figure 1B).

We next analyzed the relationship between the mRNA expression of SPAG5 and clinic characteristics, including age, sex, and grade (Table 1) in 667 glioma patients. According to the median of the SPAG5 mRNA expression, the 667 glioma specimens were further divided into low SPAG5 expression group (n = 334) and high SPAG5 expression group (n = 333). The mRNA expression levels of SPAG5 were significantly associated with overall survival time. And the high mRNA level of SPAG5 indicated poor prognosis. The expression levels of SPAG5 were not significantly associated with sex but were significantly associated with age (P < 0.001) and grade (P < 0.001). Furthermore, the overall survival between glioma patients with low (n = 334) and high (n = 333) expression of SPAG5 was compared, who were grouped according to the median of the expression of SPAG5. A Kaplan–Meier curve was obtained (Figure 1C). Glioma patients with low expression of SPAG5 showed a higher overall survival rate than glioma patients with high expression of SPAG5 [log-rank P = 0.03; hazard ratio (HR) = 3.324, CI 2.521–4.328].

The biological function of SPAG5 in glioma was next studied. Cell proliferation analysis was performed in the two cell lines (U87 and U251) transfected with SPAG5-shRNA. The mRNA and protein of SPAG5 in stable cell lines were examined by qRT-PCR and Western blotting. After transfection with SPAG5-shRNA, the mRNA and protein expression levels of SPAG5 in U87 and U251 cell lines were all downregulated (Figure 2). Knockdown of SPAG5 markedly suppressed the cell proliferation in the U87 cell lines and U251 cell lines (all P < 0.05; Figure 3). In accordance, colony formation was also significantly decreased in shSPAG5 group compared with control group on the 15th day after shRNA transfection (Figure 4). Caspase 3/7 assay was further carried out to assess the effect of SPAG5 on apoptosis (Figure 5). In the glioma cell lines (U87 and U251) transfected with SPAG5-shRNA, cell apoptosis was significantly enhanced compared with that of the normal control group. Our results revealed that SPAG5 knockdown inhibited cell proliferation and promoted apoptosis in vitro.

To further analyze the effect of SPAG5 on cell migration and invasion, wound-healing assays and invasion assays were performed. The results of wound-healing assays showed that SPAG5 depletion at 24 h inhibited the migrated cells in U87 cells (P < 0.01; Figure 6A) but had no obvious effect on U251 cells (Figure 6B). Invasion assays demonstrated that SPAG5 enhanced the ability of cell invasion in U87 and U251 cells (Figure 7). The results suggested that SPAG5 might be involved in tumor progression.

To further explore the mechanisms of SPAG5 in glioma cells, several important signaling pathway molecules were examined using Western blotting. We found that the expression of CDH2 was correlated with SPAG5 knockdown (Figures 8A, B). The results of HCS proliferation screening analysis showed that compared with the NC group, the proliferation of the KD group was significantly reduced, which was consistent with the expectation. Compared with KD group, the expression of CDH2 gene in OE group was significantly increased (Figures 8C, D). MTT showed that the proliferation of KD+NC group was decreased compared with that in NC+NC group (P < 0.05). Compared with KD+NC group, glioma cell proliferation was increased in KD+OE group (P < 0.05) (Figures 8E, F). Compared with the NC+NC group, the Transwell transfer rate in the KD+NC group decreased (P < 0.05), and the Transwell transfer rate in the KD+NC group increased (P < 0.05) (Figures 8G, H). Our results revealed that SPAG5 knockdown could reduce CDH2 expression, and overexpression of CDH2 could antagonize the effects of SPAG5 knockdown in glioma cells.