Lack of GD3 synthase (St8sia1) attenuates malignant properties of gliomas in genetically engineered mouse model

Abstract

High expression of gangliosides GD3 and GD2 is observed in human gliomas. The functions of GD3 and GD2 in malignant properties have been reported in glioma cells in vitro, but those functions have not yet been investigated in vivo. In this study, we showed that deficiency of GD3 synthase (GD3S, St8sia1) attenuated glioma progression and clinical and pathological features in a platelet-derived growth factor B-driven murine glioma model. Lack of GD3S resulted in the prolonged lifespan of glioma-bearing mice and low-grade pathology in generated gliomas. Correspondingly, they showed reduced phosphorylation levels of Akt, Erks, and Src family kinases in glioma tissues. A DNA microarray study revealed marked alteration in the expression of various genes, particularly in MMP family genes, in GD3S-deficient gliomas. Re-expression of GD3S restored expression of MMP9 in primary-cultured glioma cells. We also identified a transcription factor, Ap2α, expressed in parallel with GD3S expression, and showed that Ap2α was critical for the induction of MMP9 by transfection of its cDNA and luciferase reporter genes, and a ChIP assay. These findings suggest that GD3S enhances the progression of gliomas by enhancement of the Ap2α-MMP9 axis. This is the first report to describe the tumor-enhancing functions of GD3S in vivo.

Abbreviations
• GD3S
• GD3 synthase (ST8SIA1)
• GFAP
• glial fibrillary acidic protein
• GO
• gene ontology
• PDGFB
• platelet-derived growth factor B
• PDGFR
• platelet-derived growth factor receptor
• RCAS
• replication-competent avian leukemia virus splice acceptor
• SFK
• Src family kinase

INTRODUCTION

Gliomas are brain tumors derived from mostly glial cells.¹ Gliomas are classified into four histological grades, and the highest grade glioma, glioblastoma multiforme (grade IV), is a tumor associated with one of the poorest prognoses in adults.² Gliomas are also classified into three subtypes, mesenchymal, proneural, and classical, according to gene expression profiles.³ Loss of function of p53 and overexpression of PDGFRα have been frequently identified in the proneural subtype.

Gangliosides GD3 and GD2 are acidic glycosphingolipids, and are expressed in some types of tumors such as melanomas, gliomas, and small-cell lung cancers.^4–6 As expression of GD3 and GD2 is high and specific for these tumors, they are considered cancer-associated antigens and suitable as targets of immune therapies.⁷ Previously, we reported that ganglioside GD3 enhanced PDGFRα-mediated signals by forming a ternary complex consisting of GD3, PDGFRα, and Yes kinase in lipid rafts of the plasma membrane, leading to marked invasiveness of glioma cells in vitro.⁸ However, no studies on the roles of GD3 in actual glioma tissues in the body have been reported.

Replication-competent avian leukemia virus splice acceptor vector allows specific delivery of oncogenes to astrocyte-lineage cells using Gtv-a transgenic mice that express tv-a, an avian leukemia virus receptor, under regulation of a GFAP promoter.⁹ This is one of the murine cancer models produced by genetic engineering.¹⁰ In this system, cDNA transfection of PDGFB by RCAS into cortexes of Gtv-a mice results in the generation of gliomas in a few weeks.

In this study, we clarified how GD3 and GD2 are involved in glioma progression in vivo using an RCAS system. We compared gliomas generated in GD3S-WT mice and those in GD3S-KO mice. Here, we show that the lack of GD3S attenuated tumor properties of gliomas by modulating expression of various genes in vivo. Key molecules involved in the malignant properties of gliomas under GD3S expression were investigated.

MATERIALS AND METHODS Mice

A Gtv-a transgenic mouse that expresses tv-a under a GFAP promoter was obtained from Jackson Laboratory. The Tp53 KO mouse was obtained from Riken Bioresource Center. The GD3S-KO mouse was generated in our laboratory.¹¹ By crossing Gtv-a mice with p53 KO and GD3S KO mice, Gtv-a^+/+/Tp53^−/−/GD3S^+/+ and Gtv-a^+/+/Tp53^−/−/GD3S^−/− mice were generated. Original genetic backgrounds of these mice were: Gtv-a (129, BALB/c, FVB/N, C57BL/6), p53-KO (C57BL/6), and GD3S-KO (C57BL/6).

Generation of glioma-bearing mice

For the preparation of virus-producing cells, DF-1 cells (a chicken embryonic fibroblast cell line) were transfected with an RCAS vector harboring PDGFB-T2A-GFP cDNA by Lipofectamine 2000™ (Invitrogen), and virus-producing DF-1 cells (1 × 10⁴) were injected into the right cortexes of newborn mice by Hamilton syringe (21 G). At 7 or 14 days after injection, mice were killed with carbon dioxide gas, and brains with gliomas were harvested and imaged by IVIS Spectrum (PerkinElmer). Brains with gliomas were analyzed by H&E staining, and also used for the preparation of primary-cultured glioma cells. DF-1 cells and virus-producing DF-1 cells were maintained in DMEM supplemented with 10% FCS, penicillin, and streptomycin in a CO₂ incubator at 37°C.

Preparation of primary-cultured astrocytes and glioma cells

Brains from newborn mice or glioma tissues from mice harboring gliomas were harvested, and brains or glioma tissues were passed through a mesh (pore size, 200 µm) after removing meninges. After washing cells with PBS, they were cultured in DMEM supplemented with 10% FCS and antibiotics. To remove contaminated microglia, trypsinized cells were plated on Petri dishes and incubated for 20 minutes. Microglia attached to Petri dishes better than astrocytes. Therefore, the majority of unattached cells after incubation for 20 minutes were astrocytes. We repeated this procedure at least three times to remove microglia. To prepare primary cultured glioma cells, GFP-expressing cells were collected and enriched several times by FACS Aria II (Becton Dickinson).

Western blot analysis

Lysates prepared from tissues (50 µg) or cells (20 µg) were separated by SDS-PAGE using 8, 10, or 12% acrylamide gels. After transferring proteins onto PVDF membranes (Millipore), the membranes were blocked with 5% skim milk in PBS containing 0.05% Tween-20. Incubation with primary Abs and subsequently with HRP-labeled secondary Abs was carried out, and the membranes were then analyzed with an ECL detection system (Wako) and C-Digit Blot Scanner (M&S TechnoSystems). Antibodies for Ap2α and GFP were purchased from Santa Cruz Biotechnology. Antibodies for 2A peptide, HA epitope, and β-actin were purchased from Sigma-Aldrich. Other Abs for Akt, phosphorylated Akt (S473), Erks, phosphorylated Erks, and phosphorylated SFKs (Y417) were purchased from Cell Signaling Technology, and Abs for individual gangliosides were established in our laboratory.¹²

Microarray analysis

Pooled RNA samples of glioma tissues prepared from five each of GD3S-WT and GD3S-KO mice were analyzed by Agilent Expression Array (Agilent Technologies). The analysis of microarray results was carried out by Takara Bio.

Quantitative RT-PCR

Extraction of total RNAs from tissues and cells and reverse transcription were carried out using TRIzol reagent (Invitrogen) and M-MLV reverse transcriptase (Invitrogen), respectively, according to the manufacturer’s protocols. Complementary DNAs were analyzed by SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) and CFX Connect Real-Time PCR Detection System (Bio-Rad). The thermal cycle program recommended by Bio-Rad was used. Primers used in this study are listed in Table S1.

Flow cytometry and cell sorting

Cells were trypsinized and washed twice with ice-cold PBS. For staining of GD3 and GD2, cells were incubated with an anti-GD3 Ab (R24, provided by Dr LJ Old, Sloan-Kettering Cancer Center) or an anti-GD2 Ab (220-51)¹² in PBS for 1 hour on ice. After washing with ice-cold PBS, cells were incubated with a phycoerythrin-labeled secondary Ab in PBS for 45 minutes on ice. Expression levels of GD3 and GD2 were analyzed by FACSCalibur (Becton Dickinson), and GD3-expressing cells were collected and enriched by FACSAria II (Becton Dickinson).

Gelatin zymography

First, cells were cultured in Opti-MEM I (Thermo Fisher Scientific) supplemented with ITS (Thermo Fisher Scientific) for 2 days. Then the supernatant was collected and concentrated using Amicon Ultra-15 Filter Unit 30K (Merck Millipore). The supernatant containing 5 µg total protein was separated by SDS-PAGE including 1 mg/mL gelatin (Wako) under nonreducing conditions. After electrophoresis, the gel was washed with washing buffer (50 mmol/L Tris, pH 7.5, 2.5% v/v Triton X-100, 5 mmol/L CaCl₂, and 1 µmol/L ZnCl₂) and incubated with incubation buffer (50 mmol/L Tris, pH 7.5, 1% v/v TritonX-100, 5 mmol/L CaCl₂, and 1 µmol/L ZnCl₂) overnight at 37℃. To visualize the gelatinase activity, the gel was stained with GelCode Blue Stain Reagent (Thermo Fisher Scientific).

Invasion assay

Matrigel (Corning) was diluted five times in PBS and added on a cell culture insert (24-well plate format, 8-µm pore size; Corning) and incubated for 2 hours at room temperature to be polymerized. After washing the insert, cells (2 × 10³ in 200 µL serum-free DMEM) were plated onto the upper chamber. The lower chamber was filled with 700 µL DMEM supplemented with 10% FCS. After incubation in a CO₂ incubator at 37°C for 24 hours, invading cells were stained with Giemsa solution (Wako), and then the cell number was counted under microscopy (DMi1; Leica). Invading cells were also photographed with microscopy (BZ-X710; Keyence). For MMP9 inhibition, MMP9 inhibitor I (CAS No. 1177749-58-4; Sigma) was added into the upper chamber. The final concentration of MMP9 inhibitor I was 20 µmol/L.

Chromatin immunoprecipitation

After washing with PBS, cells (1 × 10⁷) were fixed with 1% formamide in PBS for 10 minutes and incubated with 0.125 mol/L glycine in PBS for 5 minutes. The cells were lysed by RIPA buffer (25 mmol/L Tris-HCl, pH 7.4, 150 mmol/L NaCl, 1% NP-40, 0.1% SDS, and 1% sodium deoxycholate) containing protein inhibitor cocktail (Calbiochem). The lysates were sonicated to fragment genomic DNA into approximately 600-1000 bp, and centrifuged at 20 630 g for 10 minutes to remove insoluble materials. After transferring the lysates to new tubes, immunoprecipitation was carried out with an anti-Ap2α Ab (sc-184; Santa Cruz Biotechnology) and Protein A Sepharose 4 Fast Flow (GE Healthcare). After washing three times with washing buffer (50 mmol/L Tris-HCl, pH 7.4, 1 mmol/L Na₃VO₄, 150 mmol/L NaCl, and 0.5% Triton X-100), the immunoprecipitates were heated in 500 µL Tris-EDTA at 95°C for 15 minutes. Then RNaseA (50 µg/mL; Sigma) treatment at 37°C for 30 minutes and Proteinase K (200 µg/mL; Sigma) treatment at 55°C for 1 hour were carried out. DNA purification was undertaken using phenol and chloroform by the standard method. Finally, ethanol precipitation was carried out to pellet DNAs, and the pellets were dissolved in 50 µL Tris-EDTA. Using these DNA solutions, PCR and quantitative PCR analyses were carried out. Primers used in this study are listed in Table S2.

Luciferase assay

Two micrograms each of pcDNA3.1/Ap2α tv1 or pcDNA3.1/Ap2α tv3 and pGL3-Enhancer vector harboring the MMP9 gene promoter sequence and pNL1.1TK vector (Promega) were used to transfect primary-cultured astrocytes (2 × 10⁵) using ViaFect Transfection Regent (Promega). Cells were replated in white-bottom 96-well plates (2.5 × 10⁴/well) after transfection. After culturing for 24 hours, the luminescence of the cells was measured by the Nano-Glo Dual-Luciferase Reporter Assay System (Promega) and a plate reader, Mithras LB-940 (Berthold).

Preparation of frozen sections

At 14 days after injection of PDGFB-T2A-GFP cDNA-expressing cells by RCAS, brain tissues were harvested and fixed with 4% paraformaldehyde. After incubation at 4°C overnight, tissues were subsequently incubated in 10% sucrose for 8 hours, 15% sucrose for 8 hours, and 20% sucrose overnight. Tissues were then embedded in OCT compound (Sakura Finetek). Ten micrometer-thick frozen sections of brain tissues were prepared by a cryostat (Leica).

Immunohistochemical analysis

Frozen sections were blocked with 1% BSA in PBS containing 0.075% Tween-20 for 1 hour. After washing with PBS, sections were incubated with primary Abs diluted in 1% BSA in PBS containing 0.01% Tween-20 for 1 hour. Then sections were washed with PBS and incubated with secondary Abs diluted in 1% BSA containing 0.01% Tween-20 for 1 hour. After washing with PBS, counterstaining with DAPI was carried out, and sections were mounted in 90% glycerol. Expression patterns of gangliosides were observed under confocal microscopy (FV10i-LIV; Olympus Life Science).

Study approval

All experimental protocols using mice were approved by the animal experimental committee of the Graduate School of Medicine in Nagoya University, adhering to the guidelines set by the Japanese government.

Statistical analysis

Statistical analysis was undertaken using Student’s t test. Kaplan-Meier analysis was carried out using the log-rank test.

RESULTS Lack of GD3S attenuated tumor properties and progression of gliomas

First, to clarify whether GD3 and GD2 or GD3S are involved in glioma generation and progression, we analyzed gliomas generated in the RCAS/Gtv-a glioma model. To generate gliomas, we injected PDGFB-T2A-GFP cDNA-expressing cells into right cortexes of GD3S-KO glioma model mice or GD3S-WT (control) mice by the RCAS system (Figure 1A). T2A peptide is a self-cleaving small peptide derived from Thosea asigna.¹³ The vector could express two genes linked by the T2A sequence at the same levels. By introducing this RCAS vector, gliomas expressing both PDGFB and GFP were generated. GD3S-KO mice showed a significantly longer life-span than control mice after injection (P = .0228, log-rank test) (Figure 1B). Based on the observation of GFP signals, GD3S-KO mice showed smaller gliomas than control mice at 2 weeks after injection (Figure 1C). Gliomas generated in GD3S-KO mice showed oligodendroglioma-like features, whereas those in control mice showed glioblastoma multiforme-like features in pathological analysis (Figure 1D). Immunohistochemical analysis revealed strong expression of GD3 and GD2 only in gliomas in control mice (Figures 1E and S1). More complex gangliosides than GD2 and a-series gangliosides showed similar staining patterns between gliomas and nonglioma tissues (Figure S1). These data suggest that GD3 and/or GD2 expression largely affects the nature of gliomas, and probably enhances glioma progression in vivo. Notably, in human gliomas, GD3 and GD2 expression could be enriched in the classical subtype, but there is no strong association with any subtype (Figure S2).

Lack of GD3S attenuated phosphorylation levels of Akt, Src family kinases, and Erks in glioma tissues

Next, we undertook western blot analysis using lysates prepared from gliomas generated in GD3S-KO mice and control mice at 2 weeks after the injection of PDGFB-T2A-GFP cDNA-expressing cells by RCAS. Nonglioma tissues were also prepared from the opposite site cortex of gliomas in the same mouse as the control. As shown in Figure 2, GFP and T2A expressions were restricted in gliomas. The T2A short peptide sequence worked as a tag protein for PDGFB. Therefore, anti-2A peptide Ab could detect exogenous PDGFB carrying the T2A peptide in its C-terminus. Interestingly, in glioma tissues, phosphorylation levels of Akt, SFKs, and Erks were significantly lower in GD3S-KO mice than in control mice. Generally, malignant gliomas show strong activation of signaling pathways of PI3K/Akt, SFK, and RAS/MAPK¹⁴; hence, these data indicate the lack of GD3/GD2 and/or GD3S attenuated oncogenic signals, leading to the generation of gliomas with low-grade malignancy.

Lack of GD3S resulted in alteration of gene expression profiles in gliomas

To characterize gliomas generated in GD3S-KO mice in details, we compared gene expression patterns between gliomas generated in GD3S-KO mice and those in control mice by DNA microarray. Pooled RNAs from each genetic background mouse group were used. Gene Ontology annotations for the cell cycle (GO:0007049), MAPK activity (GO:0000187), growth factor activity (GO:0008083), ECM (GO:0031012), and protein tyrosine kinase activity (GO:0004713) were carried out, and we identified a number of downregulated genes in gliomas of GD3S-KO mice compared with those of control mice by the expression levels (Figure S3) or ratios (Figure 3) between them.

Among these genes, we noted the marked downregulation of MMP family genes such as MMP3, MMP9, MMP10, MMP12, MMP13, MMP15, and MMP25 in gliomas from GD3S-KO mice. Then we analyzed gene expression levels of MMP family members in individual glioma samples generated in GD3S-KO and control mice. As shown in Figure 4A, gliomas generated in GD3S-KO mice showed low mRNA levels of MMPs compared with those in control mice, corresponding with microarray data (Figure 3). Generally, high expression levels of MMP family genes are required for cancer cell invasion. Therefore, low expression levels of MMP family genes might result in weak invasion activity of glioma cells in GD3S-KO mice.

Reintroduction of GD3S restores MMP9 expression in glioma cells

Then, we tried to identify which MMPs are upregulated by GD3S by reintroducing GD3S in gliomas generated in GD3S-KO mice. We analyzed three each of primary-cultured glioma lines prepared from control mice, GD3S-KO, and GD3S-KO cells reintroduced with GD3S (Figures 4B,C and S4). Interestingly, in the expression analysis of MMPs using these three kinds of primary-cultured glioma cells, reintroduction of GD3S in glioma cells induced restoration of MMP9 and MMP12 (Figure 4D). Actually, there was no significant difference in MMP12 expression levels between glioma tissues generated in GD3S-KO mice and control mice (Figure 4A). Consequently, only MMP9 was identified as a consistently upregulated gene by GD3/GD2 and/or GD3S in gliomas.

High expression of MMP9 contributes to strong invasion activity of glioma cells

To determine how MMP9 expression is associated with malignant tumor phenotypes, we undertook gelatin zymography and an invasion assay using primary-cultured glioma lines prepared from control mice and GD3S-KO mice. Interestingly, primary-cultured glioma cells prepared from control mice showed stronger gelatinase activity and invasion activity than those from GD3S-KO mice (Figure 5A). Additionally, the strong invasion activity of primary-cultured glioma cells prepared from control mice was attenuated by MMP9 inhibition (Figure 5B). Inhibition of MMP9 also changed the shape of invaded cells from amoeboid-like to round, which could be involved in the attenuation of invasion activity (Figure 5C). These data suggest that strong MMP9 expression induced by GD3/GD2 and/or GD3S contributes to the malignant phenotypes of glioma cells, such as marked invasiveness.

Transcription factor Ap2α upregulates MMP9 in astrocytes

To clarify mechanisms for the upregulation of MMP9 in GD3S-expressing gliomas, we focused on transcription factors that have been reported as candidate regulators of MMP9 expression, such as Ap2α¹⁵ Sp1,¹⁶ Ap1 (c-Fos, c-Jun),¹⁷ and Nfkb (p65, Nfkb2).¹⁸ Among them, Ap2α, Sp1, and p65 showed significantly higher levels in GD3⁺ astrocytes (Figure 6A). In particular, Ap2α showed markedly higher levels in GD3⁺ cells than in GD3⁻ cells. When expression levels of Ap2α were compared in glioma and nonglioma tissues from GD3S-KO and control mice, Ap2α was also expressed more strongly in gliomas in control mice than in GD3S-KO mice (Figure 6B). Ap1 (c-Fos, c-Jun) and Nfkb2 were highly expressed in gliomas in control mice compared with GD3S-KO mice. However, their expression patterns were not parallel with GD3 expression in astrocytes, as shown in Figure 6A. For this reason, we identified Ap2α as a candidate regulator of MMP9 expression under GD3 expression. We also detected higher expression levels of MMP3, MMP9, and MMP12 in GD3⁺ astrocytes than in GD3⁻ astrocytes (Figure S5).

There are two well-characterized kinds of splicing variants for murine Ap2α, tv1 (NM_011547.4) and tv3 (NM_001122948.2).¹⁹ We overexpressed these two variants to examine which Ap2α variant drives the MMP9 gene. As we expected, forced expression of Ap2α, especially Ap2α tv3, resulted in increased MMP9 gene expression in three individual primary-cultured astrocytes (Figure 7A,B). The effects of these factors on the expression of other MMP family members were also analyzed, resulting in similar but less intense effects on MMP3 and MMP10 (Figure 7C).

Transcription factor Ap2α tv3 binds and drives MMP9 promoter in astrocytes

Furthermore, we examined whether Ap2α actually binds to the MMP9 promoter and promotes expression of the MMP9 gene by ChIP-PCR and luciferase assay, respectively. Using four primer sets designed around two Ap2α binding sequences (Figure 8A), ChIP-PCR analysis was carried out and revealed evidence of binding of the transcription factor Ap2α tv3 to regions near primer sites 3 and 4 of the MMP9 gene promoter (Figure 8B). Ap2α tv1 also showed similar, but much weaker binding than Ap2α tv3. The luciferase reporter assay revealed that Ap2α tv3 drove transcription of MMP9 (Figure 8C). Taken together, activation of signaling pathways and upregulation of MMP9 induced via Ap2α based on GD3/GD2 and/or GD3S expression actually promoted the progression of gliomas.

As shown in Figure 9A, GD3S-KO mice with the introduction of GD3S cDNA together with the PDGFB expression vector tended to show a shorter survival than GD3S KO mice. In Figure 9B, the conclusion of this study is summarized. Activation of Akt, SFKs, and Erks and upregulation of MMP9 by Ap2α lead to emergence of malignant phenotypes in gliomas under the expression of GD3/GD2 and/or GD3S.

DISCUSSION

Gangliosides are localized in lipid rafts of the cell membrane and function as a fine tuner of cell signals by modulating properties of lipid rafts.²⁰ Gangliosides including GD3 and GD2 are required for neurogenesis,²¹ and play a protective role in neuroinflammation.²² In cancer cells, high and specific expression of ganglioside GD3 and/or GD2 enhances cell signals, resulting in malignant phenotypes such as rapid cell growth and increased invasiveness, as shown in small-cell lung cancer cells.²³ GD3 also enhances adhesion signals and phosphorylation levels of adaptor molecules such as p130Cas and paxillin in melanomas.^24,25 Although expression of GD3 and/or GD2 enhanced malignant properties in tumors, as reported,²⁶ there are few reports describing how GD3 and GD2 play roles in tumorigenesis and tumor progression in vivo. In this study, we found that loss of GD3S resulted in the reduction of phosphorylation levels of Akt, SFKs, and Erks in gliomas, indicating that oncogenic cell signals were attenuated due to lack of expression of GD3 and/or GD2 in gliomas (Figures 1 and 2). Interestingly, lack of GD3S dynamically altered expression profiles of a wide variety of genes in gliomas, as presented in Figure 3. Among them, we identified several MMP family genes as markedly downregulated in gliomas lacking GD3S.

There are 25 known MMPs in mammals. They are proteases targeting several kinds of proteins included in the ECM such as collagens, elastin, and gelatin. Among MMPs, MMP9 is frequently overexpressed in cancers and required for invasion of cancer cells.²⁷ Importantly, other groups reported that PI3K/AKT and MAPK pathways were essential for MMP activity as well as invasiveness in glioma cells. The association between an increased PI3K/AKT pathway and increased MMP9 activity and invasiveness was described.²⁸ Furthermore, it was also reported that activation of the Raf1-MEK-ERK pathway induces MMP9 activity and enhances invasion activity.²⁹ These studies strongly suggest that PI3K/AKT and MAPK pathways should be targeted to suppress the growth and invasiveness of glioma cells. Based on our findings, MMP9 expression levels were restored by re-expression of GD3S in GD3⁻ cells (Figure 4D), indicating that GD3/GD2-expressing gliomas have high potential for invasiveness through MMP9. Moreover, we identified the transcription factor Ap2α as a key regulator for upregulation of MMP9 (Figures 6-8). Although the expression and function of Ap2α in gliomas and other tumors have been reported, it is controversial whether Ap2α positively modulates the malignant properties of tumors.^30-32 Differential functions for each variant of Ap2α have not been well-clarified, whereas sumoylation, which occurs only in Ap2α tv1, might be involved in the regulation of Ap2α tv3 and MMP9 gene expression in gliomas.³³ Our results showing that Ap2α tv3 bound and drove the MMP9 promoter could be novel and meaningful when considering precise Ap2α functions in gliomas. However, it remains to be clarified how GD3/GD2 and/or GD3S regulate transcriptional activity of Ap2α in gliomas.

Ganglioside GD3 has been considered to be a marker of neuroprogenitor cells³⁴ due to its dominant expression in neuronal cells at E12-14 days in mice.³⁵ However, GD3S-deficient mice show almost normal development of nervous tissues,¹¹ suggesting that b-series gangliosides are not indispensable for neurogenesis. In turn, GD3 and/or GD2 have been analyzed regarding roles in various cancers, and a number of studies have reported their enhancing effects on many signaling molecules associated with cell growth, migration, invasion, and metastasis. However, many of them have been analyzed using cultured cells.²⁶ Although GD3/GD3 synthase was reported as a key molecule to maintain glioma stem cells,³⁶ the evidence was based on experiments in vitro. Therefore, the results reported here show the novel impact on the essential roles of GD3/GD2 in gliomagenesis and progression in vivo. Although major signaling pathways promoting malignant properties, such as increased cell growth and invasiveness, in glioma tissues under GD3 expression were described here and in previous reports,⁸ cross-talk between the PDGFR2α-Yes-paxillin pathway and the Ap22α-MMP9 axis remains to be clarified. AS GD3- and/or GD2-targeted therapies for neural crest-derived tumors have been long attempted and are now under development,⁷ our results could suggest novel targets for cancer therapeutics of signaling pathways activated by cancer-associated gangliosides, as we reported recently.^37,38

ACKNOWLEDGMENTS

We thank S. Yamamoto and Y. Nakayasu for technical assistance. This study was supported by Grants-in-Aids of the Ministry of Education, Culture, Sports and Technology of Japan (MEXT) (15H04696, 25670141, 24390078, 23110008, and 20K07629), and by JST-CREST (Grant Number: JPMJCR17H2).

DISCLOSURE

The authors declare no conflict of interest.