Abstract

N-Acetylgalactosaminyltransferase 2 (GALNT2), the enzyme that regulates the initial step of mucin O-glycosylation, has been reported to play a role in influencing the malignancy of various cancers. However, the mechanism through which it influences gliomas is still unknown. In the current study, the Cox proportional hazards model was used to select genes. Data obtained from The Cancer Genome Atlas (TCGA) database and immunohistochemistry (IHC) of clinical specimens showed that increased GALNT2 expression levels were associated with an unfavorable prognosis and a higher tumor grade in human gliomas. Then, GALNT2 knockdown and overexpression were performed in glioma cell lines and verified by quantitative real-time PCR (qRT-PCR) and Western blotting. Functional assays demonstrated that GALNT2 was closely related to glioma cell proliferation, cycle transition, migration and invasion. Western blot analysis and lectin pull-down assays indicated that GALNT2 knockdown decreased the level of phosphorylated epidermal growth factor receptor (EGFR) and the expression of the Tn antigen on EGFR and affected the expression levels of p21, cyclin-dependent kinase 4 (CDK4), cyclinD1, matrix metalloproteinase 2 (MMP2) and matrix metalloproteinase 9 (MMP9) through the EGFR/PI3K/Akt/mTOR pathway. GALNT2 overexpression had the opposite effects. In vivo, the growth of orthotopic glioma xenografts in nude mice was distinctly inhibited by the expression of GALNT2 shRNA, and the tumors with GALNT2 shRNA exhibited less aggressiveness and reduced expression of Ki67 and MMP2. Overall, GALNT2 facilitates the malignant characteristics of glioma by influencing the O-glycosylation and phosphorylation of EGFR and the subsequent downstream PI3K/Akt/mTOR axis. Therefore, GALNT2 may serve as a novel biomarker and a potential target for future therapy of glioma.

Introduction

Glioma, which accounts for 80% of all primary brain tumors, is the most prevalent and deadly primary tumor of the central nervous system (CNS) in adults [1,2]. Current multimodality therapy includes maximal surgical resection, radiotherapy, and chemotherapy [3]. However, the therapeutic result, which is measured by overall survival and survival quality, remains unsatisfactory [4]. Genomic alterations, such as codeletion of 1p/19q [5], methylation of the O6 methylguanine methyltransferase (MGMT) gene promoter [6,7], mutations/amplifications of the epidermal growth factor receptor (EGFR) [8], and mutations in the isocitrate dehydrogenase (IDH) gene [9], have been confirmed to be closely related to glioma tumorigenesis. Therefore, the molecular alterations that occur in glioma are expected to help predict prognosis and provide better therapeutic strategies.

N-Acetylgalactosaminyltransferase 2 (GALNT2) is the enzyme that regulates the initial step of mucin O-glycosylation [10]. Glycosylation is one of the most common event in the post-translational modification of proteins, and aberrant glycosylation can influence multiple cellular properties, including cell proliferation, transformation, differentiation, apoptosis, migration and invasion [11–13]. There are two major types of protein glycosylation: N-linked and O-linked. The mucin type is the most common type of O-glycosylation [14].

Aberrant GALNT2 expression has been reported to influence the malignancy of various cancers. For instance, GALNT2 can modify EGFR glycosylation and activity and thereby regulate the malignant behavior of HCC cells [15]. The down-regulation of GALNT2 modifies the malignant progression of gastric cancers by increasing MET phosphorylation and influencing the EGFR glycosylation and activation [16,17]. However, the function of GALNT2 has not previously been reported in glioma cells.

Increasing evidence suggests that the occurrence and development of glioma are connected to the PI3K/Akt signaling pathway [18,19]. As a tyrosine kinase receptor, EGFR is an upstream molecule in the PI3K/Akt pathway, and it becomes phosphorylated when it is activated. EGFR activation of is closely related to the glycosylation mediated by GalNAc-transferases (GALNTs) [17,20].

Here, we report that GALNT2 expression levels were associated with tumor grade and prognosis in human gliomas using The Cancer Genome Atlas (TCGA) database and immunohistochemistry (IHC) of clinical specimens. GALNT2 knockdown inhibited glioma cell proliferation, migration and invasion by suppressing the O-glycosylation and phosphorylation of EGFR and the downstream PI3K/Akt/mTOR pathway in vitro and in vivo; GALNT2 overexpression had the opposite effects. These findings indicate that GALNT2 has the potential to be a novel biomarker and a new therapeutic target in the treatment of human gliomas.

Materials and methods

Cox proportional hazards model

We used the Cox proportional hazards model to select genes that are relevant for patient’s survival and to build a predictive model for future prediction. The outcome time was defined as overall survival in months and disease-free survival in months. N genes were picked to construct the Cox regression model, for each gene Gj (j = 1, 2,…, N); we built the following Cox model for the hazard of vital status or death at time t.

where is the baseline hazard function for gene Gj, and X1, X2,…, XP are covariates. The covariates that we adjusted include race, age, sex, Karnofsky performance (KPM) score, tumor status, and history of neoadjuvant therapy.

TCGA databases and clinical specimens

The mRNA expression microarray data and the concomitant clinical information for samples were downloaded from TCGA Research Network (n=631; TCGA, http://cancergenome.nih.gov). The data were analyzed by gene set enrichment analysis (GSEA), GraphPad Prism etc. Paraffin-embedded glioma tissues (WHO II–IV) were obtained from patients (n=35) who underwent surgery at the Department of Neurosurgery, Qilu Hospital of Shandong University. Normal brain tissue samples (n=5) were collected from patients with severe traumatic brain injury who underwent partial resection of the normal brain.

IHC

Sections were obtained from paraffin-embedded tissues of normal brains and of different grades of human gliomas. The sections were heated, deparaffinized, rehydrated and placed in sodium citrate buffer (pH 6.0) for antigen retrieval, and endogenous peroxidase activity was blocked with 3% hydrogen peroxide. The slides were blocked with 10% normal goat serum and incubated with primary antibody (rabbit anti-GALNT2 monoclonal antibody, 1:50 dilution, #NBP1-83394; Novus Biologicals; U.S.A.) (mouse anti-Ki67 antibody, 1:800 dilution, #9449; Cell Signaling Technology; U.S.A.) (rabbit anti-MMP2 (matrix metalloproteinase 2) antibody, 1:150 dilution, #40994; Cell Signaling Technology; U.S.A.) at 4°C overnight. The images were visualized by following standard protocols using horseradish peroxidase–conjugated secondary antibody and 3,3′-diaminobenzidine (DAB) as the substrate. The sections were incubated with normal rabbit serum or mouse serum as negative controls. Then, the slides were counterstained with Hematoxylin, and typical images were obtained using a Leica DM 2500 microscope.

GSEA

To gain insight into the biological processes and signaling pathways associated with GALNT2 expression in gliomas, GSEA was performed using the Broad Institute GSEA version 4.0 software. The TCGA database was downloaded. The gene sets used for the enrichment analysis were downloaded from the Molecular Signatures Database (MsigDB, http://software.broadinstitute.org/gsea/index.jsp).

Cell culture

Human glioma cell lines (U87MG and U251) were obtained from the Culture Collection of the Chinese Academy of Sciences (Shanghai, China) and cultured in Dulbecco’s modified Eagle’s medium (DMEM; Thermo Fisher Scientific; U.S.A.) with 10% fetal bovine serum (FBS, Gibco; U.S.A.). These cell lines were maintained in a humidified chamber containing 5% CO2 at 37°C. SC79 (Abcam, #ab146428) was used as an Akt activator, and LY294002 (Abmole, #M1925) was used as a PI3K inhibitor.

GALNT2 knockdown and overexpression

Small interfering RNAs (siRNAs) targeting GALNT2 (GenePharma; Shanghai, China) were synthesized and transfected with Lipofectamine™ 3000 reagent (Thermo Fisher Scientific; U.S.A.) according to the manufacturer’s protocol. The knockdown efficiency was evaluated 24 h after transfection via quantitative real-time PCR (qRT-PCR) and 48 h after transfection via Western blotting. Lentiviral transduction of sh-GALNT2 (GenePharma; Shanghai, China) was used for stable knockdown of GALNT2 in cells. In parallel, for GALNT2 overexpression in glioma cells, transfection with plasmid vectors pENTER-GALNT2 or the negative control pENTER-empty (2 µg/well; both from Vigene Biosciences; U.S.A.) was performed according to the manufacturer’s protocol. The following siRNA sequences generated efficient knockdown: si-GALNT2 1: 5′-CACCCAUCAUCGAUGUCAUTT-3′; and si-GALNT2 2: 5′-GCCUUCUGCUAGAAACGUUTT-3′. The second siRNA sequence was used for the functional assays in vitro and for stable knockdown.

qRT-PCR

Total RNA was isolated from glioma cells using TRIzol reagent (Invitrogen, Life Technologies). Reverse transcription was performed using 2 μg of total RNA and the High Capacity cDNA Reverse Transcription Kits (Applied Biosystem) according to the manufacturer’s protocol. The cDNA was subjected to real-time PCR using the quantitative PCR System Mx-3000P (Stratagene). The GALNT2 primers were 5-TGTGCCTTACTGTGGTGGAC-3 and 5-GTTCCCATTTCTGTCTGCTGTC-3. The primers for GAPDH were 5-GCACCGTCAAGGCTGAGAAC-3 and 5-TGGTGAAGACGCCAGTGGA-3. The relative mRNA expression normalized to GAPDH was analyzed using GraphPad Prism 6 software.

Western blotting

Harvested cells were lysed using heat denaturation in RIPA cell lysis buffer. Protein lysates were loaded and separated using SDS/PAGE, then transferred to a polyvinylidene difluoride (PVDF) membrane. The blots were incubated with primary antibodies against GALNT2 (0.4 μg/ml; Novus Biologicals; U.S.A.), EGFR, pEGFR (Py1068) (1:5000; Abcam; U.K.), mTOR, p-mTOR, Akt, p-Akt, p21, cyclin-dependent kinase 4 (CDK4), cyclinD1, MMP2, matrix metalloproteinase 9 (MMP9), and GAPDH (1:1000; Cell Signaling Technology; U.S.A.). To visualize protein bands, enhanced chemiluminescence (ECL, Millipore, Bedford, MA, U.S.A.) was used. The intensity of protein bands was analyzed using ImageJ software and normalized to GAPDH.

Lectin pull-down assay

Vicia Villosa Lectin (VVA) agarose beads (VectorLabs, #AL-1233) were purchased to detect the Tn antigen on glycoproteins, as reported previously [17,21]. The cell lysates (0.5 mg) were incubated with 30 μl of VVA–conjugated agarose beads at 4°C for 16 h. The lectin/glycoprotein complexes were collected by centrifugation (10000 rpm, 1 min). Glycoproteins were released from the complexes after being boiled in 5 μl of 5× sample buffer for 5 min. The precipitated proteins were subjected to Western blotting to detect the amount of EGFR; the EGFR in total lysates served as the internal control [17].

Cell counting kit assay

Cell Counting Kit-8 (CCK-8) was used to measure cell viability according to the manufacturer’s instructions (Dojindo, Kumamoto, Japan). The glioma cells were seeded into 96-well plates (2 × 103 cells/200 µl/well) and cultured at 37°C, and CCK-8 solution (10 µl) was added at 24, 48 and 72 h. Following incubation for 2 h, the absorbance at 450 nm (OD450) was measured using a microplate reader (Bio-Rad).

5-ethynyl-2′-deoxyuridine cell proliferation assay

Cell proliferation rates were measured by using an 5-ethynyl-2′-deoxyuridine (EdU) cell proliferation assay kit (RiboBio, #C10310-1; Guangzhou, China) according to the manufacturer’s protocol. Glioma cells were incubated with 250 µl of EdU solution for 2 h at 37°C. Cells were fixed in 4% paraformaldehyde for 15 min, permeabilized with 0.4% Triton X-100 for 10 min, and incubated with Apollo® reagent (250 μl) for 30 min. Subsequently, the cells were stained with Hoechst 33342 for 30 min, and images were obtained using a Leica fluorescence microscope. The ratio of EdU-positive cells (Red) to total Hoechst 33342-positive cells (Blue) was used as the cell proliferation rate.

Colony-formation assay

For the colony-formation assay, cells were seeded into six-well plates at a density of 500 cells/well. DMEM containing 10% FBS was changed every third day. After 15 days, the colonies were fixed with 4% paraformaldehyde for 30 min and stained with Crystal Violet for 15 min, and representative colonies were imaged and quantitated.

Flow cytometry

Cell cycle analysis was performed using propidium iodide (PI) (BD Biosciences; U.S.A.) staining. Glioma cells were harvested, resuspended and stained with PI in the presence of RNase A for 20 min. Then, the stained cells were analyzed using a flow cytometer (BD Biosciences) according to the instructions.

3D tumor spheroid invasion assay

Glioma cells were seeded into a 3D culture qualified 96-well spheroid-formation plate (5 × 103 cells/well) and incubated in the spheroid-formation matrix (Trevigen, Gaithersburg, MD, U.S.A.) and DMEM containing 10% FBS for 72 h. When the spheroids grew to a diameter of >200 mm, the invasion matrix (Trevigen, Gaithersburg, MD, U.S.A.) was infused. The spheroids at 0 h were regarded as a reference point for measuring the area invaded by the sprouting cells. Spheroids were imaged every 24 h using a Leica microscope.

Transwell invasion and migration assays

To further assess invasiveness, filters were precoated with Matrigel. Glioma cells were added to the top chamber in serum-free medium. The bottom chamber was filled with 10% FBS DMEM. After 24 h of incubation, the top chamber cells were removed using a cotton swab, and the membrane was fixed in 4% paraformaldehyde for 15 min and stained with Crystal Violet for 15 min. Five fields of adherent cells in each well were imaged randomly. To measure migration, the filters were not precoated with Matrigel.

Wound-healing assay

Cell migration was additionally evaluated using a wound healing assay. Cells transfected with siRNA or plasmid vectors were lesioned using a plastic pipette tip, and the cells were incubated in a six-well culture cluster. Immediately and after 24 h, five randomly selected fields at the lesion border were visualized under a Leica microscope.

Intracranial mouse model

To establish intracranial gliomas, U87MG luciferase cells (5 × 105) were transfected with Lenti-sh-GALNT2 (the sequence was the same as si-GALNT2 2) or Lenti-Control virus and then stereotactically implanted into the brains of 4-week-old nude mice (SLAC Laboratory Animal Center; Shanghai, China). Bioluminescence imaging was used to detect intracranial tumor growth on days 7, 10, 14, 21 and 28. Kaplan–Meier survival curves were plotted to determine survival time and weight. Tumor tissues were harvested at 14 days after implantation, fixed in formalin, embedded in paraffin, sectioned and incubated with antibodies against GALNT2 (Novus Biologicals; U.S.A.), Ki-67 (Cell Signaling Technology; U.S.A.) and MMP2 (Cell Signaling Technology; U.S.A.).

Statistical analysis

One-way ANOVA or Student’s t test was used for data comparisons using GraphPad Prism 6 software. All data are presented as the mean ± standard error of three independent experiments. Kaplan–Meier survival curves were also analyzed via log-rank tests using GraphPad Prism 6 software. The cut-off level was set at the median value of GALNT2 expression levels. A two-tailed χ2, Fisher’s exact tests and multivariate Cox regression analysis were used to determine the association between GALNT2 expression and clinicopathological and molecular genetic characteristics by SPSS 22.0. All tests were two-sided, and P-values <0.05 were considered to be significant.

Results

High GALNT2 expression is associated with worse patient survival

The Cox proportional hazards model was used. The results showed a significant difference in the outcome for the high-GALNT2-expression group compared with the low-expression group for both overall survival and disease-free survival (Table 1). The prognostic value of GALNT2 expression in the overall survival of glioma patients acquired from TCGA was also examined via Kaplan–Meier survival curves. Patients with high GALNT2 expression (>median value) exhibited a significantly worse prognosis than those with low GALNT2 expression among all glioma patients and low-grade gliomas (LGGs) (P<0.0001). However, TCGA glioblastoma multiforme (GBM) patients analyzed separately showed no significant difference (P=0.103) (Figure 1A). Furthermore, multivariate Cox regression analysis was performed, and the results showed that GALNT2 expression was an independent indicator of overall survival (HR = 1.805, 95% CI = 1.408–2.207, P<0.001; Table 2).

GALNT2 expression is associated with patient survival and tumor grade in gliomas

Figure 1
GALNT2 expression is associated with patient survival and tumor grade in gliomas

(A) The prognostic significance of GALNT2 expression in LGG and GBM patients was analyzed in the TCGA database. The cut-off level was set at the median value of the GALNT2 levels. (B) Quantitation of GALNT2 mRNA expression levels in gliomas of different grades and subtypes in TCGA. (C) Representative images of IHC staining for GALNT2 in normal brain specimens and different grade gliomas. (D) Representative images of IHC staining for GALNT2 in specific specimens of tumor and peritumor tissue. (E) GSEA shows a positive association of high GALNT2 expression levels with cancer-related pathways, such as EMT, ECM receptor interaction, O-glycan biosynthesis, EGFR and the PI3K-Akt-mTOR signaling pathway. Magnification: ×200, upper; ×400, lower. *P<0.05; **P<0.01; ***P<0.001. Abbreviations: ECM, extracellular matrix; EMT, epithelial–mesenchymal transition; FDR, false discovery rate; NES, normalized enrichment score; ns, no statistical significance; PT, peritumor; T, tumor.

Figure 1
GALNT2 expression is associated with patient survival and tumor grade in gliomas

(A) The prognostic significance of GALNT2 expression in LGG and GBM patients was analyzed in the TCGA database. The cut-off level was set at the median value of the GALNT2 levels. (B) Quantitation of GALNT2 mRNA expression levels in gliomas of different grades and subtypes in TCGA. (C) Representative images of IHC staining for GALNT2 in normal brain specimens and different grade gliomas. (D) Representative images of IHC staining for GALNT2 in specific specimens of tumor and peritumor tissue. (E) GSEA shows a positive association of high GALNT2 expression levels with cancer-related pathways, such as EMT, ECM receptor interaction, O-glycan biosynthesis, EGFR and the PI3K-Akt-mTOR signaling pathway. Magnification: ×200, upper; ×400, lower. *P<0.05; **P<0.01; ***P<0.001. Abbreviations: ECM, extracellular matrix; EMT, epithelial–mesenchymal transition; FDR, false discovery rate; NES, normalized enrichment score; ns, no statistical significance; PT, peritumor; T, tumor.

Table 1
The Cox proportional hazards model demonstrated the effect of GALNT2 expression on outcome
GALNT2Regression coefficientP-valueOutcome
High vs. Low 0.193725005 0.009084781 Overall survival (months) 
High vs. Low 0.297642005 0.00132445 Disease-free survival (months) 
GALNT2Regression coefficientP-valueOutcome
High vs. Low 0.193725005 0.009084781 Overall survival (months) 
High vs. Low 0.297642005 0.00132445 Disease-free survival (months) 
Table 2
Univariate and multivariate Cox regression of GALNT2 expression for prognosis (overall survival) in glioma patients
VariableUnivariate Cox regressionMultivariate Cox regression
HR (95% CI)P-valueHR (95% CI)P-value
WHO grade 5.759 (3.988–8.248) <0.001 3.098 (1.879–4.950) <0.001 
High vs. Low     
GALNT2 expression 2.235 (1.902–2.503) <0.001 1.805 (1.408–2.207) <0.001 
High vs. Low     
Radiotherapy 0.435 (0.302–0.688) <0.001 0.452 (0.315–0.695) <0.001 
Yes vs. No     
IDH status 0.289 (0.187–0.396) <0.001 0.895 (0.498–1.526) 0.625 
Mutation vs. wild-type     
Age 1.041 (1.024–1.058) <0.001 0.993 (0.968–1.020) 0.352 
≥45 vs. <45     
VariableUnivariate Cox regressionMultivariate Cox regression
HR (95% CI)P-valueHR (95% CI)P-value
WHO grade 5.759 (3.988–8.248) <0.001 3.098 (1.879–4.950) <0.001 
High vs. Low     
GALNT2 expression 2.235 (1.902–2.503) <0.001 1.805 (1.408–2.207) <0.001 
High vs. Low     
Radiotherapy 0.435 (0.302–0.688) <0.001 0.452 (0.315–0.695) <0.001 
Yes vs. No     
IDH status 0.289 (0.187–0.396) <0.001 0.895 (0.498–1.526) 0.625 
Mutation vs. wild-type     
Age 1.041 (1.024–1.058) <0.001 0.993 (0.968–1.020) 0.352 
≥45 vs. <45     

Increased GALNT2 expression is associated with increased tumor grade and worse subtypes in glioma

To define the function of GALNT2 in glioma progression, the RNA expression levels of GALNT2 were analyzed in GBMs and LGGs as well as normal brain tissues from the TCGA dataset. GALNT2 mRNA levels were significantly increased in GBMs compared with those in normal brain tissues and LGGs in TCGA. GALNT2 expression was up-regulated in LGGs, according to WHO grade (Figure 1B). It has been shown that the mesenchymal subtype is associated with a worse prognosis than the proneural subtype [22]. We found that GALNT2 expression was substantially higher in the mesenchymal subtype than in the proneural subtype in the TCGA database (Table 3 and Figure 1B). We also discovered that high GALNT2 expression was significantly associated with clinicopathological characteristics of patients, such as patient age (≥45; P=0.001). Molecular genetic features, including IDH mutation, MGMT promotor methylation and codeletion of 1p/19q, have been shown to be associated with better prognosis in gliomas [23,24]. Our study demonstrated that low GALNT2 expression was associated with these features (P<0.001) (Table 3).

Table 3
Relationship between GALNT2 expression and different clinicopathological features of glioma
VariableGALNT2 high expressionGALNT2 low expressionχ2P-value
Sex   1.554 0.213 
Male 195 164   
Female 125 129   
Age   11.878 0.001 
≥45 193 136   
<45 127 157   
KPS   3.266 0.071 
≥80 155 157   
<80 42 26   
WHO grade   52.409 <0.001 
Ⅱ 71 145   
 141 96   
 108 52   
Transcriptome subtype   91.648 <0.001 
Neural 33 79   
Proneural 97 142   
Classical 65 20   
Mesenchymal 82 18   
IDH status   65.491 <0.001 
Mutant 163 263   
Wild-type 169 69   
MGMT promoter   22.244 <0.001 
Methylated 208 266   
Unmethylated 107 57   
1p/19q   61.645 <0.001 
Codeletion 40 128   
Non-codeletion 293 205   
VariableGALNT2 high expressionGALNT2 low expressionχ2P-value
Sex   1.554 0.213 
Male 195 164   
Female 125 129   
Age   11.878 0.001 
≥45 193 136   
<45 127 157   
KPS   3.266 0.071 
≥80 155 157   
<80 42 26   
WHO grade   52.409 <0.001 
Ⅱ 71 145   
 141 96   
 108 52   
Transcriptome subtype   91.648 <0.001 
Neural 33 79   
Proneural 97 142   
Classical 65 20   
Mesenchymal 82 18   
IDH status   65.491 <0.001 
Mutant 163 263   
Wild-type 169 69   
MGMT promoter   22.244 <0.001 
Methylated 208 266   
Unmethylated 107 57   
1p/19q   61.645 <0.001 
Codeletion 40 128   
Non-codeletion 293 205   

The protein levels of GALNT2 were examined via IHC in human gliomas (n=35) and normal brain tissues (n=5) from Qilu Hospital (Jinan, China). Consistent with the results from mRNA microarrays, there was obviously increased expression of GALNT2 with an increase in tumor grade (Figure 1C). Furthermore, in specific specimens, GALNT2 expression in the tumor tissue was substantially higher than that in peritumor tissue (Figure 1D). Thus, the GALNT2 expression level positively correlated with tumor grade both in the public databases and in our primary clinical specimens.

Potential biological functions and pathway analysis of GALNT2

To predict the potential biological functions and possible signaling pathway of GALNT2 in gliomas, we performed GSEA based on the GALNT2 expression in the TCGA database. The results showed that GALNT2 was concentrated in cancer-related pathways. Moreover, high GALNT2 expression was associated with the epithelial–mesenchymal transition (EMT), extracellular matrix (ECM) receptor interactions, O-glycan biosynthesis and the EGFR. In pathway analysis, we found that the PI3K/Akt/mTOR signaling pathway positively correlated with high GALNT2 expression in gliomas (Figure 1E).

Knockdown of GALNT2 induces cell cycle arrest in human glioma cells

Excessive cell proliferation and growth are characteristics of gliomas [2]. To evaluate the role of GALNT2 down-regulation in glioma cell proliferation, cells were transfected with siRNA to knockdown GALNT2, which was verified by determining the RNA expression level and protein level through qRT-PCR and Western blotting, respectively (Figure 2A,B). EdU and CCK-8 assays were performed. GALNT2 down-regulation resulted in a significant decrease in the percentage of EdU-positive cells, as well as the OD450 values, in both U87MG and U251 cells at 48 h after transfection (Figure 2C–E and Supplementary Figure S1A–C). Cell cycle analysis showed that GALNT2 down-regulation increased the population of glioma cells in the G0/G1 phase (Figure 2F,G and Supplementary Figure S1D,E). Furthermore, the colony-formation assay was used to evaluate the long-term effects of sh-GALNT2 on cell proliferation. The results showed that sh-GALNT2-transfected cells formed significantly fewer colonies than the control group (Figure 2H,I).

GALNT2 down-regulation inhibits cell proliferation and induces cell cycle arrest

Figure 2
GALNT2 down-regulation inhibits cell proliferation and induces cell cycle arrest

(A,B) qRT-PCR was used to verify RNA expression variation in U87MG and U251 cells transfected with GALNT2 siRNA or control; Western blot analysis was performed to verify protein expression; GAPDH was used as a loading control. (C,D) The EdU assay was performed 48 h after transfection (scale bar: 200 μm). (E) Growth curve based on OD450 values using the CCK-8 assay. (F,G) Cell cycle distribution was determined via flow cytometry analysis using PI staining. (H,I) The colony-formation assay was performed and analyzed in glioma cells 15 days after transfection. (J) Western blot to detect the expression levels of cell cycle regulatory factors. GAPDH was used as the loading control. The data are shown as the mean ± SEM from three independent experiments. *P<0.05; **P<0.01; ***P<0.001, relative to control. Control: nonsilencing siRNA; si-GALNT2: siRNAs targeting GALNT2; Control (H): nonsilencing shRNA; sh-GALNT2: shRNA targeting GALNT2.

Figure 2
GALNT2 down-regulation inhibits cell proliferation and induces cell cycle arrest

(A,B) qRT-PCR was used to verify RNA expression variation in U87MG and U251 cells transfected with GALNT2 siRNA or control; Western blot analysis was performed to verify protein expression; GAPDH was used as a loading control. (C,D) The EdU assay was performed 48 h after transfection (scale bar: 200 μm). (E) Growth curve based on OD450 values using the CCK-8 assay. (F,G) Cell cycle distribution was determined via flow cytometry analysis using PI staining. (H,I) The colony-formation assay was performed and analyzed in glioma cells 15 days after transfection. (J) Western blot to detect the expression levels of cell cycle regulatory factors. GAPDH was used as the loading control. The data are shown as the mean ± SEM from three independent experiments. *P<0.05; **P<0.01; ***P<0.001, relative to control. Control: nonsilencing siRNA; si-GALNT2: siRNAs targeting GALNT2; Control (H): nonsilencing shRNA; sh-GALNT2: shRNA targeting GALNT2.

In view of the GSEA results, we used Western blotting to investigate the downstream targets of GALNT2 that influence the cell cycle. GALNT2 knockdown significantly decreased the level of phosphorylated Akt and mTOR, which are essential for cancer progression in various cancers including gliomas, while the expression of total Akt and mTOR exhibited no obvious change (Figure 2J and Supplementary Figure S3A,B) [25]. In addition, CDK4 and cyclinD1 expression were decreased after GALNT2 silencing. In contrast, the cyclin-dependent kinase inhibitor p21, a tumor suppressor [26], was increased in the knockdown group (Figure 2J). In summary, these results indicated that GALNT2 down-regulation suppressed cell cycle progression in glioma cells.

Knockdown of GALNT2 inhibits glioma cell migration and invasion

GSEA implied a role for GALNT2 in the EMT process and ECM receptor interaction (Figure 1E), which play essential roles in the invasiveness of cancers [27,28]. We first performed a 3D collagen spheroid invasion assay. Silencing GALNT2 reduced the area invaded by U87MG and U251 spheroids relative to controls (Figure 3A,B and Supplementary Figure S2A,B). Next, transwell migration and invasion assays, and wound-healing assays were performed. In the transwell migration assay, the number of migrated glioma cells in the GALNT2-knockdown group was significantly decreased compared with that in the control group (Figure 3C,D and Supplementary Figure S2C,D). Transwell invasion assays revealed that compared with the control groups, GALNT2 down-regulation significantly reduced the number of glioma cells that invaded through the Matrigel-coated membrane at 48 h (Figure 3C,D and Supplementary Figure S2C,D). The wound-healing assays demonstrated that the migratory ability of glioma cells was significantly suppressed following transfection with si-GALNT2 (Figure 3E,F and Supplementary Figure S2E,F).

GALNT2 knockdown decreases the invasive ability of glioma cells

Figure 3
GALNT2 knockdown decreases the invasive ability of glioma cells

(A,B) Representative images showing invading spheroids in the 3D invasion assay for U87MG and U251 cells transfected with GALNT2 and control siRNAs evaluated at 48 and 96 h (scale bar: 200 μm). (C,D) Effect of si-GALNT2 on cell movement ability was assessed using transwell migration and Matrigel invasion assays (scale bar: 50 μm). (E,F) The wound-healing assay was used to investigate the migration capacity of the U87MG and U251 cells (scale bar: 200 μm). (G) Expression levels of the MMP2 and MMP9 proteins were detected via Western blot analysis. GAPDH was used as the loading control. The data are shown as the mean ± SEM from three independent experiments. **P<0.01; ***P<0.001, relative to control. Control: nonsilencing siRNA; si-GALNT2: siRNAs targeting GALNT2.

Figure 3
GALNT2 knockdown decreases the invasive ability of glioma cells

(A,B) Representative images showing invading spheroids in the 3D invasion assay for U87MG and U251 cells transfected with GALNT2 and control siRNAs evaluated at 48 and 96 h (scale bar: 200 μm). (C,D) Effect of si-GALNT2 on cell movement ability was assessed using transwell migration and Matrigel invasion assays (scale bar: 50 μm). (E,F) The wound-healing assay was used to investigate the migration capacity of the U87MG and U251 cells (scale bar: 200 μm). (G) Expression levels of the MMP2 and MMP9 proteins were detected via Western blot analysis. GAPDH was used as the loading control. The data are shown as the mean ± SEM from three independent experiments. **P<0.01; ***P<0.001, relative to control. Control: nonsilencing siRNA; si-GALNT2: siRNAs targeting GALNT2.

MMP2 and MMP9, members of the MMP family, play important roles in the ECM degradation, migration and invasion of tumor cells [29]. The effect of GALNT2 on the expression levels of MMP2 and MMP9 was examined; the results indicated that GALNT2 knockdown induced the down-regulation of MMP2 and MMP9 proteins in U87MG and U251 cells (Figure 3G). These results suggested that GALNT2 knockdown significantly inhibits glioma cell migration and invasion in vitro.

GALNT2 overexpression promotes glioma cell proliferation, migration and invasion

To further study the effect of GALNT2 overexpression on glioma cell proliferation, invasion and migration, cells were transfected with pENTER-GALNT2 plasmid vectors to overexpress GALNT2. Western blot analysis verified that the protein levels and the expression of p-Akt, MMP2, CDK4, and cyclin D1 were increased, while those of p21 were decreased, in the GALNT-overexpression group (Figure 4A). EdU and CCK-8 assays were also performed. GALNT2 up-regulation resulted in an increase in the percentage of EdU-positive cells and OD450 values in both U87MG and U251 cells (Figure 4B–D). Cell cycle analysis also demonstrated that GALNT2 overexpression decreased the population of the U87MG and U251 cells in the G0/G1 phase (Figure 4E,F). In transwell migration assays, the number of migrated glioma cells in the GALNT2-overexpression group was increased compared with that in the control group (Figure 4G,H). The invasion assays revealed that compared with the control groups, GALNT2 up-regulation significantly increased the number of glioma cells that invaded through the Matrigel-coated membrane at 48 h (Figure 4G,H).

GALNT2 overexpression promotes glioma cell proliferation, invasion and migration

Figure 4
GALNT2 overexpression promotes glioma cell proliferation, invasion and migration

(A) Western blotting to detect the expression of GALNT2, cell cycle regulatory factors and MMP2 with GALNT2 overexpression. GAPDH was used as the loading control. (B,C) The EdU assay was performed 48 h after transfection (scale bar: 200 μm). (D) Cell proliferation was determined using a CCK-8 assay. (E,F) Cell cycle distribution was determined via flow cytometry analysis using PI staining. (G,H) Effect of GALNT2 overexpression on cellular aggressive ability was assessed via transwell migration and Matrigel invasion assays (scale bar: 50 μm). The data are shown as the mean ± SEM from three independent experiments. *P<0.05; **P<0.01; ***P<0.001, relative to control. Vector: pENTER-empty; GALNT2: pENTER-GALNT2.

Figure 4
GALNT2 overexpression promotes glioma cell proliferation, invasion and migration

(A) Western blotting to detect the expression of GALNT2, cell cycle regulatory factors and MMP2 with GALNT2 overexpression. GAPDH was used as the loading control. (B,C) The EdU assay was performed 48 h after transfection (scale bar: 200 μm). (D) Cell proliferation was determined using a CCK-8 assay. (E,F) Cell cycle distribution was determined via flow cytometry analysis using PI staining. (G,H) Effect of GALNT2 overexpression on cellular aggressive ability was assessed via transwell migration and Matrigel invasion assays (scale bar: 50 μm). The data are shown as the mean ± SEM from three independent experiments. *P<0.05; **P<0.01; ***P<0.001, relative to control. Vector: pENTER-empty; GALNT2: pENTER-GALNT2.

Knockdown of GALNT2 decreases EGFR phosphorylation and modified EGFR O-glycosylation

Our pathway analysis from GSEA predicted a positive association of increased GALNT2 expression levels with glycolysis, O-glycan biosynthesis and EGFR in gliomas (Figure 1E). Moreover, previous studies have revealed that GALNT2 could modify EGFR O-glycosylation and influence EGFR phosphorylation in various cancers [17,20,30]. To investigate the effect of GALNT2 knockdown on EGFR phosphorylation in gliomas, the transfected cells were starved for 6 h and then stimulated with EGF (100 ng/ml) for 10 min. Our data showed that GALNT2 knockdown decreased the EGF-induced phosphorylation of EGFR in U87MG and U251 cells. However, without EGF treatment, pEGFR expression was barely detectable in U87MG and U251 cells (Figure 5A,B). To explore whether GALNT2 could influence the O-glycosylation of EGFR in gliomas, a VVA lectin pull-down assay was performed. The VVA agarose beads could detect the expression of Tn antigen (GalNAc-O-Ser/Thr) on EGFR. The result showed that GALNT2 knockdown reduced VVA binding to EGFR, which implied that GALNT2 knockdown decreased the O-glycosylation of EGFR (Figure 5C,D).

Effect of GALNT2 on the phosphorylation and O-glycosylation of EGFR in glioma cells

Figure 5
Effect of GALNT2 on the phosphorylation and O-glycosylation of EGFR in glioma cells

(A,B) GALNT2 down-regulation decreased EGF-induced phosphorylation of EGFR; pEGFR expression was normalized with that of GAPDH. (C,D) GALNT2 down-regulation decreased VVA binding to EGFR. The lysates were incubated with VVA–conjugated agarose beads. EGFR pulled down by VVA was analyzed with an anti-EGFR antibody. The expression of VVA-bound EGFR was normalized to total EGFR. The data are shown as the mean ± SEM from three independent experiments. **P<0.01; ***P<0.001. Control: nonsilencing siRNA; si-GALNT2: siRNAs targeting GALNT2.

Figure 5
Effect of GALNT2 on the phosphorylation and O-glycosylation of EGFR in glioma cells

(A,B) GALNT2 down-regulation decreased EGF-induced phosphorylation of EGFR; pEGFR expression was normalized with that of GAPDH. (C,D) GALNT2 down-regulation decreased VVA binding to EGFR. The lysates were incubated with VVA–conjugated agarose beads. EGFR pulled down by VVA was analyzed with an anti-EGFR antibody. The expression of VVA-bound EGFR was normalized to total EGFR. The data are shown as the mean ± SEM from three independent experiments. **P<0.01; ***P<0.001. Control: nonsilencing siRNA; si-GALNT2: siRNAs targeting GALNT2.

GALNT2 facilitates glioma progression through the PI3K/Akt/mTOR signaling pathway

Following the GSEA pathway analysis, we explored whether GALNT2 signals through the PI3K/Akt/mTOR pathway (Figure 1E). We treated the knockdown group with an Akt activator (SC79). As expected, the expression of p-Akt, p-mTOR, and cell cycle regulatory factors were increased, while that of p21 decreased, compared with the group without the activator (Figure 6A and Supplementary Figure S3C,D). Additionally, CCK-8 assays showed that after treatment with the Akt activator, the knockdown group exhibited an increase in proliferation ability compared with the group treated with DMSO (Figure 6B). The wound-healing assays demonstrated that the migratory ability of GALNT2-knockdown cells was also increased following treatment with the Akt activator compared with DMSO (Figure 6C,D). Next, we treated the overexpression group with a PI3K inhibitor (LY294002). The expression of p-Akt, p-mTOR, and cell cycle regulatory factors was decreased, while that of p21 increased, compared with the group without the inhibitor (Figure 6E). The results of CCK-8 and wound-healing assays showed that the proliferation and migration ability of the GALNT2-overexpression group was inhibited after treatment with the PI3K inhibitor (Figure 6F–H). These data indicated that GALNT2 delivered signals through the PI3K/Akt/mTOR signaling pathway.

Regulatory effects of GALNT2 on the PI3K/Akt signaling pathway

Figure 6
Regulatory effects of GALNT2 on the PI3K/Akt signaling pathway

(A) Cells were transfected with si-GALNT2, and the transfected cells were pretreated with the Akt activator SC79 (soluble in DMSO to 100 mM) for 24 h; the levels of the molecules downstream of Akt, cell cycle regulatory factors and MMP2 were detected via Western blotting. (B) Cell proliferation was determined using a CCK-8 assay after treatment with the Akt activator and DMSO. (C,D) The wound-healing assay was used to investigate the migration capacity after adding the activator (scale bar: 200 μm). (E) Cells were transfected with pENTER-GALNT2, and the transfected cells were pretreated with the PI3K inhibitor LY294002 (soluble in DMSO to 20 μM) for 24 h; the levels of the molecules downstream of Akt, cell cycle regulatory factors and MMP2 were detected via Western blotting. (F) Cell proliferation was determined using a CCK-8 assay. (G,H) The wound-healing assay was used to investigate the migration capacity (scale bar: 200 μm). GAPDH was used as the loading control. The data are shown as the mean ± SEM from three independent experiments. *P<0.05; **P<0.01; ***P<0.001. Control: nonsilencing siRNA; si-GALNT2: siRNAs targeting GALNT2; Vector: pENTER-empty; GALNT2: pENTER-GALNT2.

Figure 6
Regulatory effects of GALNT2 on the PI3K/Akt signaling pathway

(A) Cells were transfected with si-GALNT2, and the transfected cells were pretreated with the Akt activator SC79 (soluble in DMSO to 100 mM) for 24 h; the levels of the molecules downstream of Akt, cell cycle regulatory factors and MMP2 were detected via Western blotting. (B) Cell proliferation was determined using a CCK-8 assay after treatment with the Akt activator and DMSO. (C,D) The wound-healing assay was used to investigate the migration capacity after adding the activator (scale bar: 200 μm). (E) Cells were transfected with pENTER-GALNT2, and the transfected cells were pretreated with the PI3K inhibitor LY294002 (soluble in DMSO to 20 μM) for 24 h; the levels of the molecules downstream of Akt, cell cycle regulatory factors and MMP2 were detected via Western blotting. (F) Cell proliferation was determined using a CCK-8 assay. (G,H) The wound-healing assay was used to investigate the migration capacity (scale bar: 200 μm). GAPDH was used as the loading control. The data are shown as the mean ± SEM from three independent experiments. *P<0.05; **P<0.01; ***P<0.001. Control: nonsilencing siRNA; si-GALNT2: siRNAs targeting GALNT2; Vector: pENTER-empty; GALNT2: pENTER-GALNT2.

GALNT2 down-regulation inhibits the tumorigenesis and aggressiveness of glioma cells in vivo

To further verify the function of GALNT2 in gliomas, orthotopic xenografts were assessed in vivo. Animals bearing sh-GALNT2 cells displayed significantly reduced tumor size via bioluminescence imaging (Figure 7A). The sh-GALNT2 group exhibited a longer survival rate relative to controls (Figure 7B), and its weight loss was substantially slower than that in the control group (Figure 7C). HE staining of tumors that were collected at the same time (14 days after implantation) in the two groups also showed a smaller size in the sh-GALNT2 group, and tumors of the sh-GALNT2 cells displayed significantly more circumscribed borders (Figure 7D). IHC confirmed that GALNT2 protein levels were reduced in xenografts generated with sh-GALNT2 cells (Figure 7E). The proliferation index marker Ki-67 and the invasion marker MMP2 were also decreased in sh-GALNT2 xenografts (Figure 7F,G). These results demonstrated that GALNT2 knockdown led to reduced growth and invasion of glioma cells in vivo.

GALNT2 silencing inhibits tumorigenesis and aggressiveness in vivo

Figure 7
GALNT2 silencing inhibits tumorigenesis and aggressiveness in vivo

(A) Bioluminescence imaging showed the tumor size as days elapsed. (B) Survival analysis for animals implanted with sh-GALNT2 or control cells (P<0.05 by log-rank test). (C) Weights of two groups of mice were measured every 2 days. (D) HE staining of sections from mouse brains with control and sh-GALNT2 xenografts at 2 weeks after implantation, where the tumor size and sections at tumor margins were shown. (EG) IHC for GALNT2, Ki-67 and MMP2 in sections from the indicated xenografts. **P<0.01; ***P<0.001, relative to the control. Control: nonsilencing shRNA; sh-GALNT2: shRNAs targeting GALNT2.

Figure 7
GALNT2 silencing inhibits tumorigenesis and aggressiveness in vivo

(A) Bioluminescence imaging showed the tumor size as days elapsed. (B) Survival analysis for animals implanted with sh-GALNT2 or control cells (P<0.05 by log-rank test). (C) Weights of two groups of mice were measured every 2 days. (D) HE staining of sections from mouse brains with control and sh-GALNT2 xenografts at 2 weeks after implantation, where the tumor size and sections at tumor margins were shown. (EG) IHC for GALNT2, Ki-67 and MMP2 in sections from the indicated xenografts. **P<0.01; ***P<0.001, relative to the control. Control: nonsilencing shRNA; sh-GALNT2: shRNAs targeting GALNT2.

Discussion

At present, the prognosis of glioma patients remains very poor, even with the use of multimodal treatment strategies. Significant efforts have been made to identify prognostic molecular biomarkers that could provide knowledge regarding glioma formation and progression. In the present study, we explored GALNT2 overexpression in gliomas and demonstrated that GALNT2 expression increased as an increase in tumor grade and that GALNT2 was associated with the overall survival of patients, which suggested that GALNT2 played an important role in glioma malignancy. In support of this finding, we demonstrated for the first time that GALNT2 down-regulation inhibited glioma cell proliferation, migration and invasion by modifying the O-glycosylation and phosphorylation of EGFR and the downstream PI3K/Akt/mTOR pathway; GALNT2 up-regulation showed the opposite results.

GALNTs are a family of crucial O-glycosyltransferases that initiate the formation of mucin-type O-glycan [30]. Abnormal expression of glycosyltransferases alters the expression of glycans, which play a critical role in cancer progression [31]. GALNT2, as a member of the GALNTs, has been reported in several cancers, such as hepatocellular carcinoma [15], gastric cancer [16,17], oral squamous cell carcinoma [32] and neuroblastoma [33]. Previous studies have revealed that GALNT2 correlates with the activation of receptor tyrosine kinases (RTKs), which are important biological receptors with ligand-binding and protein kinase activity [17]. Among them, EGFR was the first to be reported and the most well-studied RTK for its roles in signal transduction and cancer progression [34]. Expression of short O-glycans has been found in many types of cancer and exploited to develop cancer vaccines [35]. Changes in these structures often alter the function of the cell and its antigenic properties, as well as its potential to invade and metastasize [35,36]. Previous studies have revealed that EGFR may express short O-glycans; however, the exact sites of O-glycosylation on EGFR were still unknown [15]. In our study, we demonstrated that GALNT2 down-regulation could inhibit EGFR activation, as indicated by the decrease in EGFR phosphorylation. Moreover, GALNT2 knockdown decreased the expression of Tn antigen on EGFR, thus influencing EGFR O-glycosylation. However, further investigation is necessary to elucidate the relationship between EGFR O-glycosylation and phosphorylation in glioma cells; we plan to investigate this relationship in future studies. The mechanism by which the activity and downstream signaling of RTKs are modified by O-glycosyltransferase might offer novel insights into the development of new therapeutics for glioma.

Abnormal cell proliferation and growth are hallmark characteristics of human gliomas [37]. Previous studies rarely reported the influence of GALNT2 on cancer cell proliferation. We performed cell proliferation assays, including EdU, CCK-8 and clonal assays; the results all demonstrated that GALNT2 down-regulation could obviously inhibit glioma cell proliferation. Flow cytometry revealed that GALNT2 knockdown induced cell cycle arrest at G0–G1. Western blot analysis indicated that GALNT2 knockdown led to significantly reduced levels of phosphorylated Akt, as well as downstream oncogenic factors including phosphorylated mTOR, cyclinD1 and CDK4, which play an important role in cell cycle G0–G1 arrest [38]. In addition, the tumor suppressor p21 was induced after GALNT2 depletion.

Previous studies have revealed that GALNT2 influenced cell migration and invasion by modifying key proteins involved in EMT [17,32,39]. The results of our pathway analysis also supported this view. We first performed a 3D tumor spheroid invasion assay, which is considered an objective assay [40]. The results showed that GALNT2 down-regulation distinctly inhibited the aggressiveness of glioma cells. Next, transwell and wound-healing assays were performed to confirm this phenomenon. Western blot analysis showed that GALNT2 knockdown decreased the expression of MMP2 and MMP9, which are considered markers of cell invasion [29]. Next, we overexpressed GALNT2; the functional assays and Western blotting showed results opposite of those obtained for the knockdown group.

Studies have shown that GALNT2 can influence the EGFR/Akt pathway [15,17], the IGF-1R pathway [33] and the MET pathway [16]. Our pathway analysis from GSEA predicted a positive association of high GALNT2 expression with EGFR and the PI3K/Akt/mTOR pathway in gliomas. Then, an Akt activator and PI3K inhibitor were used to confirm this possibility. Orthotopic xenografts of nude mice also supported the in vitro results. Notably, we found that the sh-GALNT2 tumors were obviously smaller and less aggressive than the control group. Moreover, IHC of Ki67 and MMP2 showed convincing results.

Conclusion

In conclusion, our study highlighted that increased GALNT2 expression levels were associated with an unfavorable prognosis and a higher tumor grade in human gliomas. GALNT2 facilitated glioma cell proliferation, migration and invasion by influencing EGFR O-glycosylation and phosphorylation and the downstream PI3K/Akt/mTOR pathway in vitro and in vivo. Therefore, GALNT2 may serve as a novel biomarker and a potential therapeutic target for human glioma.

Clinical perspectives

  • Glioma is the most prevalent and deadly primary tumor of the CNS in adults. Current multimodality therapies include maximal surgical resection, radiotherapy, and chemotherapy; however, the therapeutic result remains unsatisfactory. Molecular alterations that occur in glioma are expected to help predict the prognosis and provide better therapeutic strategies.

  • Our study highlighted that increased GALNT2 expression level was associated with unfavorable prognosis and a higher tumor grade in human gliomas. GALNT2 facilitated glioma cell proliferation, migration and invasion by influencing the O-glycosylation and phosphorylation of EGFR and the downstream PI3K/Akt/mTOR pathway in vitro and in vivo.

  • GALNT2 may serve as a novel biomarker and a potential therapeutic target in the treatment of human glioma. Moreover, we expect that the mechanism by which the activity and downstream signaling of RTKs are modified by O-glycosyltransferase might offer novel insights into the development of new therapeutics for glioma.

Funding

This work was supported by the National Natural Science Foundation of China [grant numbers 81571284, 91542115, 81702468, 81874083, 81802966]; the National Natural Science Foundation of Shandong Province of China [grant numbers 2017CXGC1203, 2017G006012, 2013GGE27006]; and the Taishan Scholars of Shandong Province of China [grant number ts201511093].

Author Contribution

G.L. led the study design and prepared the manuscript. Z.Z.S., H.X., Y.W. and C.C.W. performed the research and wrote the manuscript. Z.Z.S., R.Y., and J.Y.X. collected clinical samples and corresponding clinical data. Z.Z.S., Q.H.M. and S.B.W. performed cell culture and the assessment of cell functions in vitro. Z.Z.S. and M.Y.Q. conducted the in vivo work. G.L. and C.W.W. revised the manuscript. All authors read and approved the final manuscript.

Ethics Approval

We confirm that the research has been carried out in accordance with the World Medical Association Declaration of Helskini. All experimental protocols were approved by the Ethics Committee of the Qilu Hospital (Jinan, China) and performed in accordance with the relevant guidelines and regulations. Written informed consent was obtained from all patients. All animal experiments took place in the animal laboratory of the Qilu Hospital of Shandong University (Jinan, China), and were approved by the Institutional Animal Care and Use Committee (IACUC) of Shandong University (Jinan, China).

Competing Interests

The authors declare that there are no competing interests associated with the manuscript.

Abbreviations

     
  • ANOVA

    Analysis of Variance

  •  
  • CCK-8

    cell counting kit-8

  •  
  • CDK4

    cyclin-dependent kinase 4

  •  
  • CNS

    central nervous system

  •  
  • DMEM

    Dulbecco’s modified Eagle’s medium

  •  
  • ECM

    extracellular matrix

  •  
  • EdU

    5-ethynyl-2′-deoxyuridine

  •  
  • EGFR

    epidermal growth factor receptor

  •  
  • EMT

    epithelial–mesenchymal transition

  •  
  • FBS

    fetal bovine serum

  •  
  • GalNAc

    N-acetylgalactosamine

  •  
  • GALNT2

    N-Acetylgalactosaminyltransferase 2

  •  
  • GBM

    glioblastoma multiforme

  •  
  • GSEA

    gene set enrichment analysis

  •  
  • HE

    Hematoxylin-eosin staining

  •  
  • HR

    Hazard Ratio

  •  
  • IDH

    isocitrate dehydrogenase

  •  
  • IGF-1R

    Insulin-like Growth Factor-1 Receptor

  •  
  • IHC

    immunohistochemistry

  •  
  • LGG

    low-grade glioma

  •  
  • MET

    Mesenchymal to epithelial transition factor

  •  
  • MGMT

    O6 methylguanine methyltransferase

  •  
  • MMP2

    matrix metalloproteinase 2

  •  
  • MMP9

    matrix metalloproteinase 9

  •  
  • OD450

    Optical Density 450

  •  
  • PI

    propidium iodide

  •  
  • qRT-PCR

    quantitative real-time PCR

  •  
  • RIPA

    Radio Immunoprecipitation Assay

  •  
  • RTK

    receptor tyrosine kinase

  •  
  • siRNA

    small interfering RNA

  •  
  • TCGA

    The Cancer Genome Atlas

  •  
  • VVA

    Vicia Villosa Lectin

References

References
1.
Wen
P.Y.
and
Reardon
D.A.
(
2016
)
Neuro-oncology in 2015: progress in glioma diagnosis, classification and treatment
.
Nat. Rev. Neurol.
12
,
69
70
[PubMed]
2.
Omuro
A.
and
DeAngelis
L.M.
(
2013
)
Glioblastoma and other malignant gliomas: a clinical review
.
JAMA
310
,
1842
1850
[PubMed]
3.
Clarke
J.
,
Butowski
N.
and
Chang
S.
(
2010
)
Recent advances in therapy for glioblastoma
.
Arch. Neurol.
67
,
279
283
[PubMed]
4.
Louis
D.N.
,
Perry
A.
,
Reifenberger
G.
,
von Deimling
A.
,
Figarella-Branger
D.
,
Cavenee
W.K.
et al.
(
2016
)
The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary
.
Acta Neuropathol.
131
,
803
820
[PubMed]
5.
Brennan
C.W.
,
Verhaak
R.G.
,
McKenna
A.
,
Campos
B.
,
Noushmehr
H.
,
Salama
S.R.
et al.
(
2013
)
The somatic genomic landscape of glioblastoma
.
Cell
155
,
462
477
[PubMed]
6.
Hegi
M.E.
,
Diserens
A.C.
,
Gorlia
T.
,
Hamou
M.F.
,
de Tribolet
N.
,
Weller
M.
et al.
(
2005
)
MGMT gene silencing and benefit from temozolomide in glioblastoma
.
N. Engl. J. Med.
352
,
997
1003
[PubMed]
7.
Kessler
T.
,
Sahm
F.
,
Sadik
A.
,
Stichel
D.
,
Hertenstein
A.
,
Reifenberger
G.
et al.
(
2018
)
Molecular differences in IDH wildtype glioblastoma according to MGMT promoter methylation
.
Neuro Oncol.
20
,
367
379
[PubMed]
8.
Camara-Quintana
J.Q.
,
Nitta
R.T.
and
Li
G.
(
2012
)
Pathology: commonly monitored glioblastoma markers: EFGR, EGFRvIII, PTEN, and MGMT
.
Neurosurg. Clin. N. Am.
23
,
237
246
[PubMed]
9.
Zeng
A.
,
Hu
Q.
,
Liu
Y.
,
Wang
Z.
,
Cui
X.
,
Li
R.
et al.
(
2015
)
IDH1/2 mutation status combined with Ki-67 labeling index defines distinct prognostic groups in glioma
.
Oncotarget
6
,
30232
30288
[PubMed]
10.
Bennett
E.P.
,
Mandel
U.
,
Clausen
H.
,
Gerken
T.A.
,
Fritz
T.A.
and
Tabak
L.A.
(
2012
)
Control of mucin-type O-glycosylation: a classification of the polypeptide GalNAc-transferase gene family
.
Glycobiology
22
,
736
756
[PubMed]
11.
Pinho
S.S.
and
Reis
C.A.
(
2015
)
Glycosylation in cancer: mechanisms and clinical implications
.
Nat. Rev. Cancer
15
,
540
555
[PubMed]
12.
Barkeer
S.
,
Chugh
S.
,
Batra
S.K.
and
Ponnusamy
M.P.
(
2018
)
Glycosylation of cancer stem cells: function in stemness, tumorigenesis, and metastasis
.
Neoplasia
20
,
813
825
[PubMed]
13.
Dimitroff
C.J.
(
2015
)
Galectin-binding O-glycosylations as regulators of malignancy
.
Cancer Res.
75
,
3195
3202
[PubMed]
14.
Tian
E.
and
Ten Hagen
K.G.
(
2009
)
Recent insights into the biological roles of mucin-type O-glycosylation
.
Glycoconj. J.
26
,
325
334
[PubMed]
15.
Wu
Y.M.
,
Liu
C.H.
,
Hu
R.H.
,
Huang
M.J.
,
Lee
J.J.
,
Chen
C.H.
et al.
(
2011
)
Mucin glycosylating enzyme GALNT2 regulates the malignant character of hepatocellular carcinoma by modifying the EGF receptor
.
Cancer Res.
71
,
7270
7279
[PubMed]
16.
Liu
S.Y.
,
Shun
C.T.
,
Hung
K.Y.
,
Juan
H.F.
,
Hsu
C.L.
,
Huang
M.C.
et al.
(
2016
)
Mucin glycosylating enzyme GALNT2 suppresses malignancy in gastric adenocarcinoma by reducing MET phosphorylation
.
Oncotarget
7
,
11251
11262
[PubMed]
17.
Hu
W.T.
,
Yeh
C.C.
,
Liu
S.Y.
,
Huang
M.C.
and
Lai
I.R.
(
2018
)
The O-glycosylating enzyme GALNT2 suppresses the malignancy of gastric adenocarcinoma by reducing EGFR activities
.
Am. J. Cancer Res.
8
,
1739
1751
[PubMed]
18.
Lin
F.
,
de Gooijer
M.C.
,
Hanekamp
D.
,
Chandrasekaran
G.
,
Buil
L.C.
,
Thota
N.
et al.
(
2017
)
PI3K-mTOR pathway inhibition exhibits efficacy against high-grade glioma in clinically relevant mouse models
.
Clin. Cancer Res.
23
,
1286
1298
[PubMed]
19.
Lee
J.H.
,
Liu
R.
,
Li
J.
,
Zhang
C.
,
Wang
Y.
,
Cai
Q.
et al.
(
2017
)
Stabilization of phosphofructokinase 1 platelet isoform by AKT promotes tumorigenesis
.
Nat. Commun.
8
,
949
[PubMed]
20.
Park
J.H.
,
Nishidate
T.
,
Kijima
K.
,
Ohashi
T.
,
Takegawa
K.
,
Fujikane
T.
et al.
(
2010
)
Critical roles of mucin 1 glycosylation by transactivated polypeptide N-acetylgalactosaminyltransferase 6 in mammary carcinogenesis
.
Cancer Res.
70
,
2759
2769
[PubMed]
21.
Noach
I.
,
Ficko-Blean
E.
,
Pluvinage
B.
,
Stuart
C.
,
Jenkins
M.L.
,
Brochu
D.
et al.
(
2017
)
Recognition of protein-linked glycans as a determinant of peptidase activity
.
Proc. Natl. Acad. Sci. U.S.A.
114
,
E679
688
[PubMed]
22.
Arimappamagan
A.
,
Somasundaram
K.
,
Thennarasu
K.
,
Peddagangannagari
S.
,
Srinivasan
H.
,
Shailaja
B.C.
et al.
(
2013
)
A fourteen gene GBM prognostic signature identifies association of immune response pathway and mesenchymal subtype with high risk group
.
PLoS ONE
8
,
e62042
[PubMed]
23.
Yan
H.
,
Parsons
D.W.
,
Jin
G.
,
McLendon
R.
,
Rasheed
B.A.
,
Yuan
W.
et al.
(
2009
)
IDH1 and IDH2 mutations in gliomas
.
N. Engl. J. Med.
360
,
765
773
[PubMed]
24.
Jiang
T.
,
Mao
Y.
,
Ma
W.
,
Mao
Q.
,
You
Y.
,
Yang
X.
et al.
(
2016
)
CGCG clinical practice guidelines for the management of adult diffuse gliomas
.
Cancer Lett.
375
,
263
273
[PubMed]
25.
Duan
S.
,
Huang
W.
,
Liu
X.
,
Liu
X.
,
Chen
N.
,
Xu
Q.
et al.
(
2018
)
IMPDH2 promotes colorectal cancer progression through activation of the PI3K/AKT/mTOR and PI3K/AKT/FOXO1 signaling pathways
.
J. Exp. Clin. Cancer Res.
37
,
304
[PubMed]
26.
Coqueret
O.
(
2003
)
New roles for p21 and p27 cell-cycle inhibitors: a function for each cell compartment?
Trends Cell Biol.
13
,
65
70
[PubMed]
27.
Aiello
N.M.
,
Maddipati
R.
,
Norgard
R.J.
,
Balli
D.
,
Li
J.
,
Yuan
S.
et al.
(
2018
)
EMT subtype influences epithelial plasticity and mode of cell migration
.
Dev. Cell
45
,
681
695
[PubMed]
28.
Kaufhold
S.
and
Bonavida
B.
(
2014
)
Central role of Snail1 in the regulation of EMT and resistance in cancer: a target for therapeutic intervention
.
J. Exp. Clin. Cancer Res.
33
,
62
[PubMed]
29.
Andreuzzi
E.
,
Colladel
R.
,
Pellicani
R.
,
Tarticchio
G.
,
Cannizzaro
R.
,
Spessotto
P.
et al.
(
2017
)
The angiostatic molecule Multimerin 2 is processed by MMP-9 to allow sprouting angiogenesis
.
Matrix Biol.
64
,
40
53
[PubMed]
30.
Schjoldager
K.T.
,
Joshi
H.J.
,
Kong
Y.
,
Goth
C.K.
,
King
S.L.
,
Wandall
H.H.
et al.
(
2015
)
Deconstruction of O-glycosylation-GalNAc-T isoforms direct distinct subsets of the O-glycoproteome
.
EMBO Rep.
16
,
1713
1722
[PubMed]
31.
Hakomori
S.
(
2002
)
Glycosylation defining cancer malignancy: new wine in an old bottle
.
Proc. Natl. Acad. Sci. U.S.A.
99
,
10231
10233
[PubMed]
32.
Lin
M.C.
,
Huang
M.J.
,
Liu
C.H.
,
Yang
T.L.
and
Huang
M.C.
(
2014
)
GALNT2 enhances migration and invasion of oral squamous cell carcinoma by regulating EGFR glycosylation and activity
.
Oral Oncol.
50
,
478
484
[PubMed]
33.
Ho
W.L.
,
Chou
C.H.
,
Jeng
Y.M.
,
Lu
M.Y.
,
Yang
Y.L.
,
Jou
S.T.
et al.
(
2014
)
GALNT2 suppresses malignant phenotypes through IGF-1 receptor and predicts favorable prognosis in neuroblastoma
.
Oncotarget
5
,
12247
12259
[PubMed]
35.
Brockhausen
I.
(
2006
)
Mucin-type O-glycans in human colon and breast cancer: glycodynamics and functions
.
EMBO Rep.
7
,
599
604
[PubMed]
36.
Jensen
P.H.
,
Kolarich
D.
and
Packer
N.H.
(
2010
)
Mucin-type O-glycosylation putting the pieces together
.
FEBS J.
277
,
81
94
[PubMed]
37.
Hanahan
D.
and
Weinberg
R.A.
(
2011
)
Hallmarks of cancer: the next generation
.
Cell
144
,
646
674
[PubMed]
38.
Zhang
T.
,
Ji
D.
,
Wang
P.
,
Liang
D.
,
Jin
L.
,
Shi
H.
et al.
(
2017
)
The atypical protein kinase RIOK3 contributes to glioma cell proliferation/survival, migration/invasion and the AKT/mTOR signaling pathway
.
Cancer Lett.
415
,
151
163
[PubMed]
39.
Park
S.Y.
,
Kim
H.S.
,
Kim
N.H.
,
Ji
S.
,
Cha
S.Y.
,
Kang
J.G.
et al.
(
2010
)
Snail1 is stabilized by O-GlcNAc modification in hyperglycaemic condition
.
EMBO J.
29
,
3787
3796
[PubMed]
40.
Khawar
I.A.
,
Park
J.K.
,
Jung
E.S.
,
Lee
M.A.
,
Chang
S.
and
Kuh
H.J.
(
2018
)
Three dimensional mixed-cell spheroids mimic stroma-mediated chemoresistance and invasive migration in hepatocellular carcinoma
.
Neoplasia
20
,
800
812
[PubMed]

Supplementary data