TCGA (The Cancer Genome Atlas, http://cancergenome.nih.gov/) is a public repository for data storage that is freely available to users. A variety of human cancer and tumor subtype genomic mutation profiles (11), transcriptomic data (12), and clinical data (13) have been generated, providing a systematic characterization of methylation (14), miRNA expression (15), and oncogenic processes (16). Gene expression profiles of GBM were downloaded from the TCGA dataset, which contains 529 GBM samples and 10 normal samples. The data of the expression profile chip level 3 of these samples were sorted out for analyzing the differentially expressed genes (DEGs). However, multiple sets of data were assessed for certain samples in practice, thus the actual number of downloaded files was more than the original samples (548 GBM samples vs. 10 normal samples). We used P<0.05 and |FC| (fold change) ≥2 as the criteria, and the edgeR (https://bioconductor.org/packages/release/bioc/html/edgeR.html) (17,18) package in R 3.4.1 was used to identify DEGs in the GBM samples compared with normal brain samples to finally obtain the DEG list. Another gene dataset, GSE15824 (19), was downloaded from the NCBI GEO database (https://www.ncbi.nlm.nih.gov/geo/), and GPL570 was the platform file. Standardization data were carried out using the RMA algorithm of the Affy (http://bioconductor.org/packages/release/bioc/html/Affy.html) (20) package in R software, which were used for the subsequent analysis.

Database for Annotation, Visualization and Integrated Discovery (DAVID) 6.8 (http://david.abcc.ncifcrf.gov/) is an online platform that is used for gene annotation, visualization and integrated discovery (21,22). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were implemented with the DAVID database. Using this comprehensive tool, we can understand the biological meaning behind the DEGs more quickly and effectively. P<0.05 indicated a statistically significant difference.

The Search Tool for the Retrieval of Interacting Genes (STRING, http://string-db.org) database was queried to construct the protein-protein interaction (PPI) network (23). A confidence score ≥0.9 was set as the cutoff criterion, and disconnected nodes were excluded from the network. CytoHubba and Molecular Complex Detection (MCODE) in Cytoscape 3.5.1 were performed to identify hub genes and significant modules of the PPI network (24,25). The filter conditions were as follows: Degree cutoff=2, node score cutoff=0.2, k-core=2, and max. depth=100.

The GSE15824 dataset contained 40 GBM samples and 5 normal brain tissues. R software was used to evaluate the correlation between the expression of ZWINT and other related hub genes, and the Pearson correlation coefficient (R) was used to reflect the degree of linear correlations. The comparison criteria were as follows: Within −1<R<1, if R>0, indicated a positive correlation between the two genes; R<0 indicated a negative correlation, and the greater the absolute value of R, the more significant was the negative correlation; and R=0 indicated nonlinear correlation.

The human GBM cell lines U251, U87 MG, A172, and U373 were obtained from the American Type Culture Collection (ATCC). The specific origin of U87 MG cells is unknown in the ATCC version. These four cell lines were authenticated using short tandem repeat (STR) profiling analysis following ISO 9001:2008 and ISO/IEC 17025:2005 quality standards. The percent matches for U251, U373, A172 and U87 MG cells were 100, 89, 100 and 100%, respectively. They were cultured in DMEM supplemented with 10% fetal bovine serum (FBS) (Ausbian, USA) and incubated at 37°C in a humidified atmosphere of 5% CO₂.

A commercially available GBM tissue microarray (TMA) slide was purchased from Shanghai GeneChem Co., Ltd., for IHC analysis. A specific primary antibody against ZWINT was utilized for immunohistochemistry (IHC) with a 2-step protocol. IHC scores were calculated according to the intensity and proportion of positive staining. Staining intensity was classified as 0 (negative), 1 (weak), 2 (moderate) and 3 (strong). The positive cell ratio of the stained cells was scored as 0 (<1%), 1 (1–25%), 2 (26–50%), 3 (51–75%) and 4 (76–100%). The multiplied result of the two scores represented the protein levels of ZWINT, and an immunoreactive score (IRS) >6 was defined as a high expression level of ZWINT, while an IRS ≤6 indicated a low level of ZWINT.

Total RNA was extracted from cells using a TRIzol kit (cat. no. 3101-100, Pufei Biotechnology), and complementary DNA (cDNA) was synthesized with the M-MLV reagent (Promega). The SYBR Master Mix (Takara) and Mx3000P Real-Time PCR machine were used to perform RT-qPCR. The PCR primers (GeneChem) were ZWINT-F, 5′-CACGTAGAGGCCATCAAAATTGG-3′ and ZWINT-R, 5′-CGGAGTTGTGTCCGTTTCCT-3′; GAPDH-F, 5′-TGACTTCAACAGCGACACCCA-3′ and GAPDH-R, 5′-CACCCTGTTGCTGTAGCCAAA-3′. Relative quantification was calculated using the ΔΔCq method (26), normalized based on GAPDH, and conducted in triplicate.

According to the cDNA sequence of ZWINT (Gene Bank accession no. NM_007057) and the design principle of the small hairpin RNA (shRNA), three target characteristic sequences and nonspecific sequences were selected. shRNA against ZWINT was transfected into 293T cells using Lipofectamine 2000™ reagent (Invitrogen; Thermo Fisher Scientific, Inc.), and the expression of ZWINT at the protein level was detected by Western blot analylsis to determine the transfection efficiency. The optimal target sequence was selected: 5′-TTCTCCGAACGTGTCACGT-3′. ZWINT shRNA and the control shRNA were provided by GeneChem Co., Ltd. ZWINT shRNA was used to knock down ZWINT expression in the U251 and U87 MG cells with Lipofectamine 2000 reagent. At 72 h after transfection, U251 and U87 cells were harvested for later research.

293T cells were lysed in radioimmunoprecipitation assay (RIPA) buffer after transfection with shRNA for 72 h. A BCA Protein Assay Kit was used to detect the total protein concentration. Equal amounts of protein lysates were electrophoresed in 10% SDS-PAGE gels and then transferred to polyvinylidene difluoride (PVDF) membranes, which were blocked by Tris-buffered saline with Tween (TBST) containing 5% milk. Protein expression was probed with mouse anti-FLAG (F1804; Sigma-Aldrich; Merck KGaA) and goat anti-mouse IgG (SC-2005; Santa Cruz Biotechnology, Inc.). Protein expression in 293T cells that were exogenously transfected with a plasmid encoding FLAG-tagged ZWINT was detected by mouse anti-FLAG and mouse anti-GAPDH (SC-32233; Santa Cruz Biotechnology, Inc.). The ECL-PLUS kit (Thermo Fisher Scientific. Inc) was used to detect the signals on X-ray film. GAPDH was used as the reference protein to calculate the relative protein levels. Independent tests were performed in triplicate.

The number of cells at each time point was counted by a Celigo Image Cytometer (Nexcelom Bioscience). Briefly, the GBM cells were cultured in 96-well plates at an initial density of 2×10³ cells/well. Each group had three wells (10 µl/well) and was incubated at 37°C in an atmosphere of 5% CO₂. Fluorescence photomicrographs were captured, and cells with green fluorescence were measured by a Celigo Image Cytometer. Cell growth curves were generated for a time course of 5 days. MTT [3- (4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] (Genview Scientific. Inc.) was used to measure cell viability. The infected cells (2×10³ cells/well) were collected and reseeded in 96-well plates. From the first to the 5th day, a total of 20 µl MTT solution (5 mg/ml) was added to the cells. Subsequently, the supernatants were removed, 100 µl DMSO was added to each well, and the plates were oscillated for 3 min. Finally, the optical density (OD) at 490 nm was measured with a microplate reader.

U251 and U87 MG cells were transfected with ZWINT shRNA or NC, cultured for 72 h and harvested. The invasion assay, in which the Transwell chamber was coated with Matrigel, was conducted according to the manufacturer's instructions for the Corning Invasion Kit (Corning Inc.). A total of 500 µl cell suspension in FBS-free DMEM (10⁵ cells/well in a 24-well plate) was added to the upper chamber, and 750 µl DMEM containing 30% FBS was added to the lower chamber. Cells were removed from the upper chamber of the filter with a cotton swab after 24 h of incubation at 37°C. Cells on the underside were washed with PBS, stained with Giemsa, captured with a digital camera and counted in 5 randomly selected fields of vision at ×100 magnification and 9 randomly fields of vision at ×200 magnification under a phase contrast microscope.

Apoptotic cells were assessed using the Annexin V apoptosis kit (88–8007; eBioscience). In short, cells were trypsinized, washed and centrifuged at 283 × g for 5 min. Then, 1X binding buffer was used to wash the cells again by resuspending the cells in 200 µl. Then, the cells were incubated with 10 µl Annexin V-APC at room temperature for 15 min in the dark. Finally, the Annexin V-stained cells were analyzed with a flow cytometer (Millipore) to determine the proportion of apoptotic cells.

Caspase-3/7 are central effector caspases in apoptosis and are usually used to measure apoptotic activities. GBM cells were first transfected with the ZWINT shRNA and NC. Next, 1×10⁴ infected cells/well were seeded in a 96-well plate. Caspase-Glo 3/7 reagent (100 µl, G8091; Promega) was added to each well, the plate was shaken for 30 min constantly and incubation was carried out at ambient temperature for 2 h. The luminescence signal was detected with an M2009PR (Tecan Infinite) plate reader.

Animal experiments were approved by the Animal Care Committee of Tongji Hospital of Huazhong University of Science and Technology (Wuhan, Hubei, China), and strictly followed the institutional regulations and state guidelines on experimental animals. BALB/c athymic female nude mice (4-week-old; ~20 g) were purchased from Shanghai Lingchang Biological Polytron Technologies Inc. They were maintained in a constant temperature, humidity, sterile environment and fed with food and water according to the national regulations. Subcutaneous xenografts of human GBM were established by injecting 5×10⁶ ZWINT shRNA U87 MG cells or NC cells into the right hind limbs of the mice; 10 mice in each group. A digital caliper was used to measure tumor growth once every 3 days. Thirty-six days after cell inoculation, the animals were euthanized using an intraperitoneal injection of 150 mg/kg pentobarbital sodium before collecting the tumors. Tumor tissues were excised and weighted. The tumor volume was defined as V≈π/6 × L × W², where L stands for the tumor length and W is the tumor width.

SPSS version 20.0 (IBM Corp.) was used for data analysis. Measurement data are expressed as the mean ± standard deviation. Student's t-test was used to compare data between two groups. The Pearson correlation coefficient was used for the correlation analysis. P<0.05 was considered to indicate a statistically significant difference.

The downloaded gene expression profiles in the TCGA database included 2,387 upregulated genes and 3,873 downregulated genes. The GO functional analysis demonstrated that the upregulated genes were mainly involved in biological processes associated with mitotic nuclear division, cell division, sister chromatid cohesion, protein binding and cell proliferation, and the ZWINT gene was mainly enriched in cell division, mitotic sister chromatid segregation, kinetochore, and mitotic cell cycle checkpoint. The downregulated genes were mainly enriched in positive regulation of excitatory postsynaptic potential, neurotransmitter transport, axon terminus, and ion channel binding. Moreover, five significant signaling pathways were overexpressed in the upregulated genes, including the cell cycle, cellular senescence, DNA replication, Ras and MAPK signaling pathways. Only one KEGG pathway was found for downregulated genes when P=0.05 (Table I).

The PPI network of DEGs included 1,843 nodes and 5,116 edges, with each node representing a DEG and the edges indicating the interactions between DEGs. Twenty genes were selected as hub genes, which included tumor protein p53 (TP53), polo like kinase 1 (PLK1), nuclear division cycle 80 (NDC80) and Wnt family member 5A (WNT5A). Moreover, ZWINT interacted with kinetochore protein NDC80 homolog (NDC80), serine/threonine-protein kinase PLK1 (PLK1), spindle and kinetochore associated complex subunit 1 (SKA1) and tripartite motif containing 17 (Terf/TRIM17) (Fig. 1A). The significant module included 9 nodes and 36 edges (Fig. 1B). GO and KEGG pathway enrichment analyses revealed that genes in this module, such as NDC80, SKA1 and NUF2 component of NDC80 kinetochore complex (NUF2), were mainly associated with cell division, mitotic cell cycle, chromosome segregation, and small GTPase-mediated signal transduction, and they interacted with ZWINT.

Correlation analysis of the public dataset GSE15824 showed that ZWINT mRNA expression was significantly positively correlated with NDC80, PLK1, NUF2, SKA1, SPC24 component of NDC80 kinetochore complex (SPC24), SPC25 component of NDC80 kinetochore complex (SPC25), shugoshin 1 (SGOL1) and shugoshin 2 (SGOL2) expression, and the correlation coefficients were R=0.929, R=0.879, R=0.945, R=0.876, R=0.888, R=0.965, R=0.804, and R=0.939, respectively. According to the biological functions of ZWINT, which was enriched in GBM, these 8 hub genes may also be correlated with cell division and mitotic cell cycle (Fig. 2).

Our data confirmed that ZWINT was among the upregulated genes, and the expression level of ZWINT mRNA in GBM tissues was significantly higher than that in normal brain tissues (548 GBM samples vs. 10 normal samples, Fig. 3A). In addition, IHC analysis of the TMA was performed to examine the protein level of ZWINT, and a significant increase was identified in GBM tissues compared with the level noted in the paired normal brain tissues (Fig. 3B). Scoring results showed that ZWINT was overexpressed in GBM tissues, and the difference was statistically significant (P<0.05) (Fig. 3D). qPCR analysis revealed that ZWINT was upregulated in all four human GBM cell lines (Fig. 3C). Therefore, U251 and U87 MG cells were selected for ZWINT-knockdown experiments. Western blot analysis revealed that in the 293T cell line transfected with shRNA targeting ZWINT (shZWINT), the expression level of ZWINT protein was significantly reduced, indicating an effective lentivirus-delivered shRNA sequence (Fig. 4A). Over 80% of U251 and U87 MG cells showed green fluorescent protein expression under a fluorescence microscope after infection with the recombinant lentiviruses (Fig. 4B). As detected by qPCR, a significant reduction in the mRNA level was found in U251 and U87 MG cells infected with shZWINT compared with cells infected with shCtrl (Fig. 4C).

The proliferation activity in U251 and U87 MG ZWINT-knockdown cells was explored by MTT assays. The results showed that the growth rate of shZWINT U251 and U87 MG cells was slower than that of the shCtrl group (P<0.01, Fig. 5A). Similarly, the growth curve counted by the Celigo Image Cytometer revealed that clonogenic survival in the shZWINT group was markedly decreased (Fig. 5B). Therefore, we speculated that ZWINT may act as an oncogene to promote GBM cell proliferation.

Transwell assays were performed to investigate the role of ZWINT in GBM cell invasion. As expected, the number of cells in the shZWINT group that passed through the membrane into the lower chamber was significantly lower than that in the shCtrl group. The results showed that there were fewer invading cells in the shZWINT group than that in the control group at 24 h after invasion, and the invasive ability was significantly inhibited (P<0.01, Fig. 6).

In FACS analysis with Annexin-V, significantly increased apoptosis was found in the ZWINT shRNA-infected U251 and U87 MG cells from 1.57 to 11.91% (P<0.01) (Fig. 7A). Similarly, as shown in caspase-3/7 activity analysis, the level of caspase-3/7 activity in the cells expressing shZWINT was significantly higher than that in cells expressing shCtrl (P<0.01) (Fig. 7B). Overall, ZWINT may have an important role in inhibiting cell apoptosis.

To further demonstrate the role of ZWINT in vivo, we performed tumor xenograft assays. The average growth rate of the tumors in the shZWINT mouse group was significantly slower than that of the shCtrl group, and the shZWINT group had significantly decreased tumor volumes and weights compared to those of the shCtrl group (Fig. 8). The results indicated that ZWINT may act as a potent protumorigenic factor that can accelerate GBM growth.