The 121 glioma tissue samples were randomly obtained from the Department of tissue, Xiangya Hospital between January 2009 and December 2012. All the gliomas were pathology confirmed by 2 independent pathologists according to the 2007 WHO classification. All patients eligible for this study acceptted surgery and were regularly followed-up, and collected detailed clinicopathological data and survival data. Magnetic resonance imaging (MRI) or contrast-enhanced MRI was performed every 6 months. OS was defined as the time from the surgery to the death or the last follow-up visit. PFS was defined as the time from the surgery to the first evidence of recurrence, progression, or death. All patients signed a written informed consent. The study was approved by the ethics committee of Xiangya Hospital in accordance with the Declaration of Helsinki.

The human primary glioma cell line U87 was obtained from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China), and cultured in DMEM (Gibco, NY, USA) with 10% fetal bovine serum (Gibco) at 37°C in a humidified atmosphere of 5% CO₂.

Total RNA of glioma tissues and cells were isolated with TRIzol reagent (Invitrogen, CA, USA) following the manufacturer's instructions. The concentration and purity of all RNA samples were determined by NanoDrop Spectrophotometer (NanoDrop Technologies, TX, USA). Then, the cDNA was synthesized by reverse transcription of total RNA using cDNA synthesis kit (Roche Life Sciences, Switzerland), and real-time PCR reaction was conducted using the SYBR Green PCR kit (Roche Life Sciences) and performed on ABI PRISM 7100 Sequence Detection system (Applied Biosystems, CA, USA) following the manufacturer's instructions. GAPDH served as control. The target mRNA expression was calculated using the 2^−∆∆Ct method. Each experiment was repeated 3 times. The primers were as follows: BAF53a (forward, 5′-CCAGGTCTCTATGGCAGTGTAA-3′; reverse, 5′-CGTAAGGTGACAAAAGGAAGGTA-3′); GAPDH (forward, 5′-GTCTCCTCTGACTTCAACAGCG-3′; reverse, 5′-ACCACCCTGTTGCTGTAGCCAA-3′); E-cadherin (forward, 5′-GCCTCCTGAAAAGAGAGTGGAAG-3′; reverse, 5′-TGGCAGTGTCTCTCCAAATCCG-3′); vimentin (forward, 5′-AGGCAAAGCAGGAGTCCACTGA-3′; reverse, 5′-ATCTGGCGTTCCAGGGACTCAT-3′).

Fresh tissues or cells were dissolved by RIPA lysis buffer (Invitrogen) and PMSF (Zhongshan Goldenbridge Biotechnology, Shanghai, China). Then isolated by centrifugation and quantified by Bradford Protein assay kit (Beyotime Institute of Biotechnology, Shanghai, China). Next the protein was separated by 10% SDS-PAGE and transferred to the PVDF membrane (Sigma, MO, USA). The membranes were blocked with 5% calf serum and incubated with appropriate primary antibody, followed by incubation with secondary antibody. Lastly, the signal was detected using enhanced chemiluminescence regents (Thermo Scientific, MA, USA) and photographed by camera and image processing system (Bio-Rad, CA, USA).

The glioma tissues were embedded in paraffin, and sectioned into slides with a thickness of 4 µm. Then, sections were dewaxed in xylene and rehydrated in graded ethanol. Antigen retrieval was accomplished using citrate buffer at pH 6.0. Then 3% H₂O₂ was used to block endogenous peroxidase activity, and goat serum incubation was used to block the non-specific antigen binding, followed by incubation with appropriate concentration of primary antibody at 4°C overnight. Then, the sections were incubated with the corresponding secondary antibody for 30 min at room temperature after rinsing three times in PBS. The antigen-antibody interactions was detected by DAB and counterstained by hematoxylin, followed by dehydrated in graded ethanols and mounted. The results were evaluated by two independent pathologists blinded to the study. The staining intensity of BAF53a was scored according to the percentage of positive stained tumor cells, as negative (−) <5% of tumor cells stained positive, (+) 0–25%, (++) 26–75%, >76% of tumor cells stained positive.

Full-length human BAF53a overexpression clone lentivirus and short hairpin RNAs (shRNA) lentivirus and their control vectors were constructed by a biotechnology company (GeneChem, Shanghai, China). Cells were cultured in 6-well plates before transfection until 80–90% confluency within 24 h. Then the lentivirus were transfected into cells according to the manufacturer's instructions. Puromycin (2 µg/ml) was used to select stable clones if necessary. After 48–72-h transfection, the cells were harvested for subsequent assays. Transfection efficiency was assessed by fluorescence microscope, real-time PCR and western blotting.

Mitomycin-C and puromycin were purchased from ApexBio, Technology LLC A4452, A3740; ApexBio, TX, USA). The Matrigel matrix was purchased from Corning Life Sciences (354248; Corning, MA, USA). The primary antibodies against BAF53a (sc-137062), E-cadherin (sc-71007), vimentin (sc-80975), and β-actin (sc-130300) were purchased from Santa Cruz Biotechnology (CA, USA).

The proliferation potential was measured by MTT assay and colony formation assay. For MTT assays, the cells were grown into 96-well plates with a density of 5×10³ cells/well. Each day 6 wells of each group were added 100 µl MTT (0.5 mg/ml; Sigma) reagent per well and incubated at 37°C for 4 h. Subsequently, the absorbance values were measured at 570 nm. For colony formation assays, 2×10³ cells/well were seeded into 6-well plates and cultured for 14 days. Then the cell colonies were dyed with crystal violet and counted. The results were recorded as mean ± SD of three repeats.

The migration potential was evaluated by wound-healing and Transwell assay. The cells were cultured into 6-well plates at a density of 2×10⁵ cells/well, when grew to 90% confluence, cells were incubated with mitomycin-C (10 µg/ml) for 1 h to suppress proliferation, and starved in serum-free medium for 24 h. When cells were confluent monolayer artificial wound was scraped by a 10-µl pipette tip. The migration gap of the wound was assessed after 24 or 48 h. The migration potential was evaluated by Transwell system. The cells treated with mitomycin-C (10 µg/ml) for 1 h at 37°C in serum-free medium were placed into them with a density of 1×10⁵ cells/insert. The DMEM with 10% FBS was added into the lower chamber. After incubation for 24–48 h, the cells in the upper membrane of insert were removed and the cells adhering to the lower membrane of the inserts were stained with crystal violet and counted by microscope after rinsing. Then the invasion potential was evaluated by Transwell-Matrigel system. The culture upper inserts were coated with Matrigel, the subsequence processes were carried out as Transwell assay. All the experiments were performed in triplicates.

Statistical analysis was conducted with SPSS version 18.0 (SPSS, Inc., Chicago, IL, USA). Student's two-tailed t-test was used to determine the statistical significance of measurement data. Survival curves were performed by the Kaplan-Meier method and compared by the log-rank test. Univariate and multivariate Cox regression model was used for survival analysis. The continuous data are shown as the mean ± SEM. A P-value of <0.05 was considered statistically significant.

To examine the role of BAF53a in glioma, we initially detected the protein expression of BAF53a in tumor tissues by immunohistochemistry (IHC). BAF53a was predominantly expressed in the nucleus. We defined high BAF53a expression as strongly (+++) or moderately (++) positive staining, and low BAF53a expression as weakly positive (+) or negative staining (Fig. 1A). BAF53a was positively expressed in the majority of glioma tissues (102/121), of which 79 exhibited high expression (Table I). Additionally, BAF53a mRNA expression was examined in 10 randomly selected cases of glioma and 10 normal brain tissues obtained from resection following trauma. These results indicated that BAF53a mRNA expression levels in glioma tissues were significantly higher than those in normal brain tissues (P<0.01; Fig. 1B).

Subsequently, the expression profile of BAF53a was explored using the Oncomine™ database (https://www.oncomine.org). The results showed that the gene copy number of BAF53a in astrocytoma was higher than that in normal brain tissue (Sun brain: fold change, 2.460; P<0.01; Fig. 1C). Furthermore, the BAF53a gene copy number in glioblastoma was significantly higher than that in normal brain tissue (TCGA brain: fold change, 11.006; P<0.01; Murat brain: fold change, 5.298; P=0.002; Fig. 1D). These results enable us to speculate that BAF53 may be involved in the progression of glioma.

We next explored the detailed associations between BAF53a protein expression (determined by IHC) and the clinicopathological features and prognosis of glioma patients. The results indicated that high BAF53a expression was significantly correlated with WHO grade (P=0.031; Table I). Survival analysis showed that the BAF53a low expression group had favorable overall survival (OS) [median OS time, 23.0 (95% CI, 11.6–34.4) vs. 12.0 (95% CI, 8.9–15.1) months; P<0.001; Fig. 2A] and progression-free survival (PFS) [median PFS time, 18.7 (95% CI, 10.6–26.8) vs. 8.0 (95% CI, 6.7–9.3) months, P<0.001; Fig. 2B] compared with the BAF53a high expression group. Furthermore, univariate and multivariate Cox regression analyses indicated that BAF53a expression was an independent prognostic factor for the OS and PFS of glioma patients (HR=1.752, P=0.013 and HR=1.874, P 0.008, respectively; Tables II and III). These results indicate that BAF53a may be useful as a prognostic marker for patients with glioma.

In order to study the functional role of BAF53a in glioma, stable BAF53a-overexpressing U87^BAF53a and BAF53a-knockdown U87^shBAF53a cells and control cells transfected with empty vector were established. The transfection efficiency in each cell type was confirmed by qPCR and western blotting (Fig. 3A and B). Subsequently, the cells were subjected to MTT and colony formation assays. The MTT assay revealed that BAF53a overexpression could markedly increase U87 cell growth, while BAF53a knockdown had the opposite effect as compared with control cells (P<0.05, respectively; Fig. 3C). The colony formation assay also showed that the U87^BAF53a cells were significantly larger and greater in number than the control cells, while U87^shBAF53a cells exhibited the opposite characteristics (P<0.05, respectively; Fig. 3D). These results indicated that BAF53a could promote the proliferation of glioma cells.

Wound healing and Transwell assays were used to assess the migratory ability of U87 cells expressing varying levels of BAF53a. The results showed that BAF53a-overexpressing U87^BAF53a cells exhibited increased migration and faster wound healing capacities, whereas U87^shBAF53a cells exhibited decreased migration and slower wound healing capacities when compared with control cells (Fig. 3E and F). A Transwell-Matrigel chamber assay accordingly showed that U87^BAF53a cells had a greater capacity for invasion across the Matrigel membrane compared with the control cells, whereas U87^shBAF53a cells had reduced invasive ability (Fig. 3G). These results indicated that BAF53a could promote the invasion capacity of glioma cells and potentially facilitate the metastasis of glioma.

As previous studies have reported a possible link between BAF53a and EMT in other physiological processes and cancer progression, we aimed to explore this association in glioma. The expression levels of two EMT markers, E-cadherin and vimentin, were detected in cells expressing different levels of BAF53a. The qPCR and western blotting results showed decreased expression of the epithelial marker E-cadherin and increased expression of the mesenchymal marker vimentin in U87^BAF53a cells. By contrast, the opposite pattern of expression was demonstrated in U87shBAF53a cells (Fig. 4A and B). Alterations in cellular morphology assessed by phase contrast microscopy also revealed that U87^BAF53a cells had more pseudopodia and were elongated, indicating a mesenchymal phenotype, whereas U87^shBAF53a cells had an oval or cobblestone-like appearance (Fig. 4C). E-cadherin and vimentin expression were also detected in glioma tissues by IHC. The results revealed that E-cadherin expression was frequently absent in glioma, whereas vimentin was consistently expressed (Fig. 4D). The Spearman rank correlation coefficient indicated that BAF53a expression was positively correlated with the expression of vimentin (r=0.384, P<0.001) and negatively correlated with the expression of E-cadherin (r=−0.367, P<0.001) (Fig. 4E). Taken together, these results suggest that BAF53a may promote glioma progression via EMT.