Sustained elevation of Snail promotes glial-mesenchymal transition after irradiation in malignant glioma

Abstract

Background

Ionizing irradiation is an effective treatment for malignant glioma (MG); however, a higher rate of recurrence with more aggressive phenotypes is a vital issue. Although epithelial-mesenchymal transition (EMT) is involved in irradiation-induced cancer progression, the role for such phenotypic transition in MG remains unknown.

Methods

To investigate the mechanism of irradiation-dependent tumor progression in MG, we performed immunohistochemistry (IHC) and qRT-PCR using primary and recurrent MG specimens, MG cell lines, and primary culture cells of MG. siRNA technique was used for MG cell lines.

Results

In 22 cases of clinically recurrent MG, the expression of the mesenchymal markers vimentin and CD44 was found to be increased by IHC. In paired identical MG of 7 patients, the expression of collagen, MMPs, and YKL-40 were also elevated in the recurrent MGs, suggesting the The Cancer Genome Atlas-based mesenchymal subtype. Among EMT regulators, sustained elevation of Snail was observed in MG cells at 21 days after irradiation. Cells exhibited an upregulation of migration, invasion, numbers of focal adhesion, and MMP-2 production, and all of these mesenchymal features were abrogated by Snail knockdown. Intriguingly, phosphorylation of ERK1/2 and GSK-3β were increased after irradiation in a Snail-dependent manner, and TGF-β was elevated in both fibroblasts and macrophages but not in MG cells after irradiation. It was noteworthy that irradiated cells also expressed stemness features such as SOX2 expression and tumor-forming potential in vivo.

Conclusions

We here propose a novel concept of glial-mesenchymal transition after irradiation in which the sustained Snail expression plays an essential role.

Malignant glioma (MG) frequently occurs in the adult brain and is one of the most aggressive neoplasms among the human cancers.¹ Therapeutic modalities in general include surgical resection and fractionated radiotherapy as well as concomitant and adjuvant chemotherapy with alkylating drugs such as temozolomide.² Nonetheless, the incidence of recurrence, regrowth, and dissemination is remarkable, resulting in a high mortality rate and poor prognosis. Recent genotyping and expression profiling analyses have shown that MG can be categorized into 4 subtypes: proneural, neural, classical, and mesenchymal based on The Cancer Genome Atlas (TCGA) study.^3–5 The proneural subtype, which shows high expression of the genes implicated in neurogenesis, is associated with better clinical outcome, especially with IDH-1 mutation and PDGFPA expression. In contrast, the mesenchymal subtypes are characterized by more aggressive phenotypes, presumably due to high expression of genes related to cellular proliferation and angiogenesis.⁶ It has also been reported that MG frequently shifts towards the mesenchymal subclass upon recurrence,⁷ although the underlying molecular mechanisms have not yet been elucidated.

Epithelial-mesenchymal transition (EMT) was originally described as a critical mechanism in embryonic development induced by a range of intrinsic and extrinsic factors including transforming growth factor (TGF)-β,⁸ epidermal growth factor (EGF),⁹ hepatocyte growth factor (HGF),¹⁰ and various other cytokines.¹¹ EMT elicits mesenchymal change in epithelial cells, followed by increased motility through the transcriptional regulators for EMT such as Slug, Snail, and Twist. All of these transcription factors are indispensable for embryonic development, and they play a spatiotemporally distinct role during embryonic development.^12,13 Several studies have shown that EMT is also related to wound healing,¹⁴ tissue remodeling,¹⁵ cancer invasion,¹⁶ cancer motility,¹⁷ stemness,^18,19 and tumor survival after irradiation.²⁰

In MG cell lines, the capabilities of motility and invasion have been ascribed to Slug,²¹ Snail,²² Twist,²³ Zinc finger E-box-binding homeobox (ZEB)1,²⁴ and ZEB2 expression.²⁵ Slug and Twist have also been expressed within the mesenchymal area of gliosarcoma specimens,^26,27 suggesting the possibility that mesenchymal transition-like EMT might contribute to so-called “postirradiation malignant progression of MG”. Meanwhile, it has been shown that sublethal irradiation promotes migration and invasion of cells through the TGF-β,²⁸ vascular endothelial growth factor (VEGF), and EGF pathways in MG.²⁹

In this study, we found that the expression levels of mesenchymal markers including vimentin, fibronectin, α-SMA, collagen, and matrix-metalloproteinase (MMP) (which were related to EMT) and CD44 and YKL-40 (which were related to the mesenchymal subtype based on TCGA study) were increased in clinically recurrent MG. In addition, we identified Snail as the master regulator of irradiation induced glial-mesenchymal transition (GMT) possibly through the phosphorylation of GSK-3β and extracellular signal-regulated kinase (ERK)1/2, resulting in the promoted migration and invasion. TGF-β, which was released from microenvironment, may also accelerate GMT after irradiation.

Materials and Methods

Cases

This study was approved by the Medical Ethics Committee of Hokkaido University Graduate School of Medicine. Surgically resected primary tumors of MG were diagnosed according to the World Health Organization (WHO) classification. The patients included in this study had undergone irradiation therapy after surgical resection and exhibited local recurrences (Supplementary Table 1 and Table 2). Twenty-two paired identical tumors were analyzed by immunohistochemistry (IHC), and 7 paired identical tumors were analyzed by quantitative RT-PCR (qRT-PCR). (Supplementary Table 3).

Immunohistochemistry

Resected tumors from patients were fixed in formalin, dehydrated with increasing concentrations of ethanol, and then impregnated with and embedded in paraffin. Primary antibodies for GFAP, vimentin, α-SMA, Olig2, PDGFRA, EGFR, and CD44 were used. The expression of these molecules was assessed according to proportion score and intensity score by 3 pathologists (RM, MT, ST). Proportion score was defined as follows: 0 (<1%), 1 (1%–10%), 2 (11%–50%), and 3 (51%–100%), and intensity score was defined as follows: 1 (weakly positive), 2 (moderately positive), and 3 (strongly positive). Total score was represented by the sum of the proportion scores and intensity scores. To exclude false positivity of each marker protein, as in the inflammatory cells, we performed IHC of CD3, CD20, CD68, and CD34 and confirmed that the analyzed area of the tumors did not include a significant amount of inflammatory cells (data not shown).

Cell Culture

Two human malignant glioma cell lines were used: T98G (ATCC#CRL-1690; American Type Culture Collection) and KMG4 (kindly provided by Dr. Kazuo Tabuchi of Saga University, Saga, Japan). Both cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM) (Wako) supplemented with 10% fetal bovine serum (FBS), 1% penicillin, and 1% streptomycin. Cell cultures were maintained in a humidified atmosphere of 5% CO₂ at 37°C. For zymography and TGF-β1 quantification, cells were maintained in DMEM without FBS for 24 hours prior to sample preparation. To assess the effect of TGF-β stimulation on the phosphorylation of Smad2/3, recombinant TGF-β1 (Sigma) was added to the medium at a final concentration of 5 ng/mL. We also established primary patient-derived malignant glioma cell lines. These cells were maintained in astrocyte basal medium (Lonza), supplemented with FBS, L-glutamate, ascorbic acid, rhEGF, insulin, and gentamicin sulphate/amphotericin B according to the manufacturer's directions, and kept in a humidified atmosphere of 5% CO₂ at 37°C. For co-culture experiments, glioma cells were seeded at a concentration of 5 × 10⁴ cells per mL of DMEM and allowed to adhere to the dish. Twenty-four hours later, murine leukemic monocytes/macrophage cell line RAW 264.7 cells (ATCC#TIB-71) or murine fibroblast NIH3T3 (ATCC#CRL-1658) were seeded at a concentration of 1 × 10⁴.

Irradiation

Cells were irradiated either with a non-fractionated single dose of 10Gy or multifractionated 20Gy divided into 2Gy doses on each weekday for 2 weeks. All experiments were done at room temperature using the LINAC III accelerator (6MV X-ray, Siemens) at a rate of 2.5Gy/minute. Control cells were placed in the irradiation room without exposure to radiation.

Reagents

Recombinant human TGFβ1, MEK1/2 inhibitor, U0126, and GSK-3 inhibitor SB 216763 were purchased from Sigma, and diluted in distilled water or DMSO according to the manufacturer's directions. U0126 and SB 216763 were added to the medium 24 hours prior to irradiation.

RNA Extraction and RT-PCR

Total RNA was extracted using the RNeasy Mini Kit (Qiagen), where 1 μg of RNA was subjected to reverse transcription (RT) with the Superscript VILO (Invitrogen). The resulting cDNA was utilized for semi-qRT-PCR and qRT-PCR using GoTaq Green Master Mix (Promega) and SYBR Green DNA polymerase (Applied Biosystems), respectively. The sequences of primers are provided in Supplementary Tables 4 and 5. The following protocols were optimized for semi-qPCR for SOX2, OCT3/4, Nanog, and GAPDH. Amplification of SOX 2 was performed for 35 cycles (30 s at 95°C, 30 s at 55°C, and 1 min at 72°C), that of OCT3/4 was 38 cycles (30 s at 95°C, 30 s at 68°C, and 1 min at 72°C), that of Nanog was 33 cycles (30 s at 95°C, 30 s at 55°C, and 1 min at 72°C), and that of GAPDH was 34 cycles (30 s at 95°C, 30 s at 65°C, and 1 min at 72°C).

Immunoblotting

Immunoblotting was performed using the method previously described.² Cells were washed twice in cold PBS and lysed in a buffer containing 0.5% NP-40, 10 mM Tris-HCl (pH7.4), 150 mM NaCl, 1 mM EDTA, 50 mM NaF, 1 mM PMFS, and 1 mM Na₃VO₄, and the lysate was clarified by centrifugation at 15,000 rpm for 15 minutes. The supernatants were size-fractionated by SDS-PAGE; 15% SDS-PAGE was used for Slug and Snail, while 7% SDS-PAGE was used for tubulin. The separated proteins were transferred to polyvinylidene difluoride membranes and blocked using Tris-buffered saline (TBS) containing 5% skimmed milk at room temperature for 1 hour. Primary antibodies were incubated at 4°C overnight and then washed with TBS, followed by incubation with secondary antibodies. The membranes were washed 3 times in PBS, and signal was developed using SuperSignal West Femto Maximum Sensitivity Substrate Solution (ThermoFisher Scientific), followed by detection using the LAS1000 (Fujifilm). Antibodies used were monoclonal rabbit anti-Slug, monoclonal anti-Snail, anti-phospho-Smad2, anti-Smad2, anti-phospho-Smad 3, anti-Smad 3, anti-phospho-GSK-3β, anti-phospho- ERK1/2, anti-ERK1/2, anti-phospho-AKT, anti-AKT, anti-phospho-STAT3, anti-STAT3 (all from Cell Signaling Technology), anti-GSK-3β, (BD Transduction), and anti-Tubulin (Sigma Aldrich) antibodies.

siRNA and Transfection

Among 4 candidates of siRNA targeting human Snail that were purchased from Qiagen, 2 siRNA (siRNA#1 and #2) with higher knockdown efficacy were utilized in this study. The sequences of these siRNA were as follows: siSnail#1 (si1), GAGGTGTGACTAACTATGCAA; siSnail#2 (si2), CCGAATGTCCCTGCTCCACAA. Seventy-two hours prior to irradiation, 5 × 10⁶ cells were transfected with 600 ng of siSnail using HiPerfect transfection reagent (Qiagen).

Snail Expression Plasmid

Full-length cDNA for human Snail was obtained by reverse transcription-PCR (RT-PCR) of total RNA from human 293T cells and cloned into pCR2.1-Topo-TA (Invitrogen). To generate Flag-tagged Snail expression vector, the cDNA was subcloned into the Not1 and XhoI sites of a pCXN2-Flag expression vector. T98G cells were transfected with pCXN2-Flag-Snail using Fugene HD reagent (Promega), and Snail expression was examined by RT-PCR and immunoblotting using anti-Flag (Sigma Aldrich) and anti-Snail Abs (Cell Signaling Technology).

Chromatin Immunoprecipitation Assay

A Chromatin Immunoprecipitation (ChIP) Assay Kit (Millipore) was used according to the manufacturer's directions. Briefly, control and Flag-tagged Snail-overexpressing glioma cells were fixed with 1% formaldehyde, lysed in lysis buffer, sonicated, and clarified by centrifugation, and the resulting supernatant was immunoprecipitated with anti-Flag antibody. The precipitates were subjected to PCR to amplify the promoter regions in MMP-2³⁰ and SOX2.³¹ The PCR products were resolved electrophoretically using 2% agarose gel and visualized by ethidium bromide staining. It was imaged using the LAS1000 (Fujifilm).

Immunofluorescence

Cells were seeded at a density of 1 × 10⁵ in 35 mm glass bottom dishes (IWAKI) and cultured overnight. The cells were then fixed in 3% paraformaldehyde for 15 minutes, permeabilized with 0.1% Triton X-100 for 4 minutes, and blocked with 1% BSA for 20 minutes. Anti-paxillin antibody (Ab) (BD Transduction) and anti-SOX2 antibody (Cell Signaling Technology) were incubated overnight at 4°C, and the secondary Ab conjugated with AlexaFluor488 (Invitrogen) was incubated for 1 hour at room temperature. F-actin was visualized by phalloidin conjugated with AlexaFluor594. Images were acquired using a confocal laser-scanning microscope (FV-300, Olympus). The relative SOX2 expression of each condition was obtained by dividing the SOX2 expressing cell number by the number of cells expressing SOX2 in control.

Wound-healing Assay

Cells were seeded at a density of 5 × 10⁴ in a 35 mm dish to achieve a 90% confluence 24 hours later. Cells grown to confluence were scratched off using a pipette tip. The cells were washed twice with warmed PBS and incubated in DMEM without FBS for 24 hours and then photographed. Cell migration was calculated by measuring the distance covered by migrating cells and further dividing by that of the original wound (% of recovery). The relative wound closure was evaluated as the ratio to % of recovery of the control cells without irradiation and was represented as relative motility index in graph as previously described.³²

Transwell Migration Assay

The uncoated Boyden chamber for transwell assay (8 μm pore size, Corning Costar) was used to quantify chemotactic migration. 5 × 10⁴ cells suspended in serum-free medium were seeded into the upper chamber, DMEM with 10% FBS was added in the lower chamber, and after 12 hours the remaining cells were removed from the upper chamber by wiping. The cells that had passed through to the lower side of the filter were fixed with 100% methanol and stained with 0.04% crystal violet. The total number of migrating cells was quantified by counting the amount of cells per 5 random 400x fields.

Matrigel Invasion Assay

Following the manufacturer's protocol, the chambers were allowed to defrost at room temperature and rehydrated in DMEM. 5 × 10⁴ cells were then seeded in the upper well and incubated for 24 hours. The remaining cells in the upper well were removed by wiping. The cells moved to the lower side of the filter were fixed with 100% methanol and stained with 0.04% crystal violet. The total number of invading cells was quantified by counting in 5 random fields (400 × high power field). To evaluate the invasive potential of cells, invasion index was calculated as follows: the number of invading test cells/the number of invading control cells.³³ This experiment was performed in triplicate.

Colony Formation Assay

Following the protocol established by Watanabe et al,³⁴ 1 × 10⁶ cells were seeded in 0.4% soft agar and incubated for 7 days and 14 days for KMG4 and T98G cells, respectively, in a humidified atmosphere of 5% CO₂ at 37°C. The number of colonies was determined by microscopy.

Gelatin Zymography

Cells were washed with PBS twice and cultured in serum-free DMEM for 24 hours, and conditioned medium was clarified by centrifugation at 1,500 rpm for 15 minutes. 20 μL of conditioned medium was mixed with an equal volume of 2X non-denaturing loading buffer and size-fractionated by 10% SDS-PAGE containing 25 mg/mL gelatin. The gels were washed twice in 2.5% Triton X-100 in PBS for 15 minutes at room temperature to remove SDS, rinsed twice in water, and incubated at 37°C for 24 hours in an incubation buffer (50 mM Tris, 5 mM CaCl₂.H₂0, 1 mM ZnCl₂). The resulting gel was stained and fixed with 50% methanol and 10% acetic acid containing 1.25 mg/mL Coomassie Brilliant Blue for 30 minutes and then destained by 10% methanol and 5% acetic acid. The zymogram was imaged using the LAS1000, and intensities of bands, which correspond to gelatinase activity, were quantified using Multigauge software (Fujifilm).

ELISA for TGF-β1

The levels of secreted TGF-β1 in the conditioned medium were analyzed using the Human/Mouse TGF beta 1 ELISA Ready-SET- Go! (eBioscience), according to the manufacturer's instructions.

In Vivo Experiments

All in vivo experiments were sanctioned by the Hokkaido University Ethics Committee for animal experiments. KMG4 cells were transfected with siControl (siC) or siSnail2 (si2) at 48 hours prior to 10 Gy irradiation. After 21 days, 1 × 10⁴, 1 × 10³, and 1 × 10² cells in PBS were inoculated subcutaneously into 6-week-old female NOD/SCID mice for xenograft experimentation. For an orthotopic model of brain tumor, 1 × 10² KMG4 cells in same setting were injected into the right forebrain of 8-week-old female NOD/SCID mice. Body weight, size of tumor mass, and general condition were monitored twice a week, and mice were euthanized when tumors reached 2 cm in maximum diameter or the animal was in obvious distress.

Statistical Analysis

The data are presented as means and standard deviations (SD) from 3 independent experiments; each was performed in triplicate. Statistical comparisons between 2 groups were performed using the parametric Student t test. For comparison of more than 3 groups, one-way analysis of variance (ANOVA), followed by Tukey's multiple comparison, was performed. Survival curves were generated according to follow-up data with the Kaplan–Meier method, and comparison between cumulative survival rates was performed using log-rank test. In all cases, P < .05 was considered statistically significant.

Results

Expression Levels of the Mesenchymal Markers Vimentin, α-SMA, and CD44 Were Increased in Human Recurrent Malignant Glioma Specimens

Histopathological analyses of human MG specimens by IHC demonstrated that expression levels of the mesenchymal markers vimentin and α-SMA increased in recurrent MG compared with primary tumor, whereas that of the glial marker GFAP was reduced (Fig. 1A–C, Supplementary Fig. 1 and Table 1). By individually scoring the results of IHC, we confirmed the statistical significance of the increase of vimentin (Fig. 1C). In addition, qRT-PCR analysis demonstrated that the other typical mesenchymal markers such as collagen (COL)1A1-3, α-SMA, fibronectin, and MMP-9 were clearly elevated in recurrent tumors (Fig. 1D). Considering the TCGA-based mesenchymal subtype of MG, the total score of CD44 was significantly increased, and Olig2 was decreased in recurrent glioma compared with the primary tumor. However, the expression levels of PDGFRA and EGFR were not significantly changed in IHC (Supplementary Fig. 1A and 1B, Supplementary Table 2). qRT-PCR analysis confirmed the elevation of mesenchymal markers such as CD44 and YKL-40, which were observed in recurrent gliomas (Fig. 1E).

Fig. 1.

Irradiation induces upregulation of mesenchymal markers in 22 patients with malignant glioma (MG). (A–E) H&E staining and immunohistochemistry for GFAP, vimentin, and α-SMA were performed using surgically resected primary and recurrent tumors from 22 MG patients. The pictures of a representative area are demonstrated. The bar indicates 20 μm. (B) To determine the total score, tumor sections were graded based on surface area and intensity of the staining, as described in Materials and Methods. Total score is represented as the sum of surface area and intensity scores. (C) Total score transitions in primary versus recurrent tumors of individual patients are shown in graphs. The width of lines indicates the incidence for the corresponding patients, and the elevation and decrease of total score in the recurrent tumors are shown as red and blue colors, respectively. Black represents unchanged. “Pri” and “Rec” represent primary and recurrent tumors, respectively. (D and E) mRNA was extracted from primary and recurrent tumors. mRNA levels of EMT-related mesenchymal markers (D) and representative markers in the TCGA subgrouping (E) were determined by qRT-PCR. Expression levels are shown as relative expression to that of primary tumors. *P < .05.

Irradiation Increased the Expression Levels of Mesenchymal Markers and EMT Regulators in T98G and KMG4 MG Cells

To ensure the postirradiated increase of mesenchymal markers observed in the clinical specimens, we employed 2 MG cell lines (T98G and KMG4) that were irradiated and used for the further analysis. First, we categorized the T98G and KMG4 cell lines based on the TCGA classification, where KMG4 with PDGFRA (+) and T98G with PDGFR (-) and EGFR (+) could be classified into the proneural and classical subtypes, respectively, according to the previously reported algorithm³⁵ (Supplementary Fig. 2). In this study, cells were irradiated with a single dose (10 Gy) or with a multifractionated regimen of 20 Gy divided into 10 fractions (1-day interval), and the expression levels of marker proteins were examined at 48 hours and 21 days after irradiation. In these cells, postirradiated elevation of EMT-related mesenchymal markers COL1A2 and/or a-SMA (Fig. 2A) and TCGA-based mesenchymal markers, including CD44 and YKL-40, could be observed in irradiated T98G and KMG4 cells (Fig. 2B).

Fig. 2.

Irradiation increases the EMT- and TCGA- related mesenchymal markers and EMT regulators in the glioblastoma cell lines T98G and KMG4. (A and B) Cells were treated with or without fractionated irradiation (20 Gy at 2 Gy/d). At 21 days after irradiation, mRNA levels of EMT-related (A) and TCGA-based (B) mesenchymal markers were examined by qRT-PCR and are shown as relative expression to control. (C and D) Cells were irradiated at single dose (10 Gy) and fractionated irradiation (20 Gy at 2 Gy/d). mRNAs were extracted at 48 hours and 21 days after irradiation, and the levels of Slug, Snail, and Twist mRNA were determined by qRT-PCR (C) along with immunoblotting using antibodies to the indicated proteins (D). (E) mRNA levels of EMT regulators in 7 primary and recurrent tumors were determined by qRT-PCR. Expression levels are shown as relative expression to that of primary tumors. *P < .05; †P < .001; **P < .0001.

To investigate the mechanism of mesenchymal transition, the roles of EMT regulators were analyzed at 48 hours and 21 days after irradiation. Although all of the EMT regulators including Slug, Snail, and Twist tended to increase at each time point in both cell lines, only Snail mRNA had a sustained increase after single and multifractionated irradiation (Fig. 2C). Similar results, including Snail elevation, were obtained in the analyses of protein levels both at 48 hours and 21 days (Fig. 2D). To ensure the significance of the clinical correlation between irradiation and upregulation of mesenchymal markers based on EMT and TCGA subtype, qRT-PCR analysis was performed using primary and recurrent tumors of 7 patients, and Snail mRNA was significantly increased in the recurrent tumors compared with primary tumor (Fig. 2E).

Irradiation Induced Increase of Mesenchymal Markers and EMT Marker Snail Were Observed in Primary Cultured Human Malignant Glioma Cells

For confirmation, 3 MG patient-derived primary cultured cells G121, G135, and G140 were prepared, and qRT-PCR analysis was performed. We found significantly increased levels of EMT-related mesenchymal markers Col1A1 and/or Col1A2 and/or α-SMA mRNA and mesenchymal markers based on TCGA study, CD44 and/or YKL-40 mRNA at 21 days post irradiation (Fig. 3A and B), together with an increased level of Snail among the 3 EMT regulators 21 days after irradiation (Fig. 3C).

Fig. 3.

Irradiation induces the expression of EMT- and TCGA-based mesenchymal markers and EMT regulators in glioblastoma patient-derived primary cell lines. Three patient-derived primary cell lines (G121, G135, and G140) were treated with and without fractionated irradiation (20 Gy at 2 Gy/d). At 21 days after irradiation, mRNA levels of EMT-related mesenchymal markers (A), TCGA subtype markers (B), and EMT-regulators such as Snail, Slug, and Twist (C) were examined by qRT-PCR. *P < .05.

Snail Knockdown Suppresses Irradiation-dependent Col1A1, Col1A2, and α-SMA Expression at 21 Days After Irradiation

To verify the significance of Snail in irradiation-induced EMT-like phenomenon, Snail mRNA was knocked down by siRNA in T98G cells, and 2 distinct siRNAs (si1 and si2) significantly repressed the mRNA levels of Snail (Supplementary Fig. 3A). Of note, this depletion of Snail gave rise to decreased mRNA of other transcription factors such as Slug and Twist (Supplementary Fig. 3A). In our system, the attenuation of Snail was sustained until 21 days after irradiation (Fig. 4A and B, Supplementary Fig. 4A and B) in both mRNA and protein levels. Decreased Snail expression in the protein level at both 48 hours and 21 days is consistent with Snail mRNA suppression. At 21 days post irradiation, a significant increase was detected in the mRNA levels of Col1A1, Col1A2, and α-SMA (Fig. 4C). In the context of Snail knockdown, irradiation-evoked mRNA expression of Col1A1, Col1A2, and α-SMA was significantly suppressed in T98G (Fig. 4C) and KMG4 (Supplementary Fig. 4C). The same results were also reproduced in patient-derived primary glioma cells G121, G135, and G140 (Fig. 4D and E for G121, data not shown for G135 and G140).

Fig. 4.

Knockdown of Snail suppresses irradiation-induced expression of EMT-related mesenchymal markers. T98G (A–C) and patient-derived primary glioblastoma cells G121 (D and E) were transfected with 2 different Snail-targeting siRNA (#1; si1, #2; si2) or a control siRNA (siC), followed by irradiation or no irradiation, respectively. Levels of Snail mRNA (A and D) and protein (B) after 48 hours and 21 days were determined and are shown as relative expression to that of control cells. mRNA levels of Col1A1, Col1A2, and α-SMA at 21 days after irradiation were quantified by qRT-PCR relative to control in T98G (C) and G121 cells (E). The data represent mean of independent experiments performed in triplicate. *P < .05.

For confirmation of the roles for Snail in the mesenchymal phenotype, we established Snail overexpressing T98G cell lines and examined the expression levels of EMT regulators and mesenchymal markers. The overexpression of Snail significantly increased the mRNA levels of Slug and Twist (Supplementary Fig. 3B) as well as MMP-2 and MMP-9 (Supplementary Fig. 3C), indicating the significance of Snail.

Snail Promotes Cell Motility and Invasion of Glioblastoma via MMP-2 Activity

Because Snail knockdown suppressed the expression of irradiation-induced mesenchymal molecules, we next investigated the impact of Snail expression on cell motility and invasion. Immunofluorescence analysis of control cells (0 Gy with siC) revealed that paxillin, a major component of focal adhesion, was localized at the end of thin actin fibers throughout the control cells (Fig. 5A, left upper panel). In the irradiated cells without siRNA (10 Gy with siC), the cells were more elongated and stretched. Alarge number of enlarged focal adhesions, as visualized by paxillin, were localized at the end of shorter actin stress fibers in one direction, demonstrating that marked remodeling of focal adhesions and actin cytoskeletons took place (Fig. 5A, right upper panel). In the presence of siSnail (10 Gy with si1, 10 Gy with si2), the cell morphology seemed to be restored to the characteristics of control cells (Fig. 5A, lower panels).

Fig. 5.

Knockdown of Snail suppresses irradiation-evoked cell motility, invasion, proliferation, MMP-2 activity, and prolonged survival of mice. (A–E, G and H) T98G cells were transfected with 2 different Snail-targeting siRNA (#1; si1, #2; si2) or a control siRNA (siC), followed by irradiation or no irradiation, respectively. The cells at 21 days after irradiation were subjected to respective assays as described in Materials and Methods. (A) Immunofluorescence analysis was performed with an antibody to paxillin (green) and phalloidin for F-actin (red). (B) In wound healing assay, relative wound closure was calculated and shown in graph. *P < .05. (C) In transwell migration assay, the number of migrated cells on the bottom side of the filter after 12 hours was counted, and the number of migrated cells was graphed. *P < .05. (D) In matrigel invasion assay, the number of invaded cells on the bottom side of the matrigel-coated filter after 24 hours was counted, and relative invasion index was graphed, as described in Material and Methods. Values are means ± SD from 3 independent experiments. *P < .05. (E) Activity of MMP-2 in conditioned media from indicated cells was shown in gelatine zymograph (upper panel), and the relative activity was graphed (lower panel). (F) Snail-overexpressing-KMG4 and -T98G cells were subjected to ChIP assay, and Snail-dependent promoter activity of MMP-2 was determined. (G) In colony formation assay, 1 × 10⁶ cells were seeded in 0.4% soft agar and incubated for 14 days, after which the number of colonies in the indicated size range was determined by microscopy. *P < .05, (H) KMG4 cells were transfected with Snail-targeting siRNA (#2; si2) or a control siRNA (siC), followed by irradiation. For an orthotopic model of brain tumor, 1 × 10² cells were injected into the right forebrain of NOD/SCID mice. Kaplan–Meier survival is shown in graph.

In the wound-healing assay, the rate of wound closure was increased after irradiation, which was abrogated by siSnail (Fig. 5B). In the chemotaxis assay using a transwell chamber, the cell motility after irradiation was significantly increased in comparison with the control, whereas the downregulation of Snail by siRNA cancelled this elevation (Fig. 5C and Supplementary Fig. 5A). Matrigel invasion assay for evaluating the invasive potential provided similar results: upregulation after irradiation and downregulation by Snail knocked-down (Fig. 5D and Supplementary Fig. 5B). We also found that the gelatinase activity of matrix metalloproteinase (MMP)-2 was increased in cells after irradiation, and again this effect was also abolished by siSnail transfection (Fig. 5E). This Snail-mediated MMP-2 expression was also confirmed by ChIP assay using the MMP2 promoter region (Fig. 5F). Furthermore, irradiated cells have the potential to form significantly larger-sized colonies in soft gel compared with control cells, and Snail knockdown cells significantly decreased the total number of colonies, especially those with large sizes (Fig. 5G, Supplementary Fig. 5C, and 5D). In all the assays mentioned above, the inhibitory effect by si2 was greater than that by si1 (Fig. 5B–E and G), which corresponded to a downregulated efficacy of Snail per se (Fig. 4A and B).

To assess the effect of Snail on tumor formation and the survival in vivo, 1 × 10² KMG4 cells post irradiation were orthotopically implanted into the right forebrain of NOD/SCID mice. It was noteworthy that the mice injected with KMG4 cells depleted of Snail did not die during the 60 days of observation, whereas those injected with the corresponding control cells died at the ratio of 60% (Fig. 5H, Supplementary Fig. 5E).

Irradiation Increases Stemness Gene Expression in Malignant Glioma Cell Lines

Because stem cell population has been implicated in treatment resistance for several cancers, we examined whether the irradiation-resistant MG cells acquired stem cell-like features. Semi-qPCR for stemness genes revealed that the expressions of Sox2, Oct3/4, and Nanog were remarkably increased after irradiation and that these increases were clearly canceled by the knockdown of Snail (Fig. 6A, Supplementary Fig. 6A). Immunofluorescence analysis revealed that KMG4 cells also showed increased expression levels of Sox2 in the nucleus after irradiation (Fig. 6B, Supplementary Fig. 6B, right upper panel), and the effect was canceled in Snail knockdown (Fig. 6B, Supplementary Fig. 6B, lower panels). Relative Sox2 expression was almost twice as high as control after irradiation, whereas downregulation of Snail diminished this elevation. The statistical analysis demonstrated significant increase and decrease of Sox2 expression by irradiation and Snail knockdown, respectively (Fig. 6C). The similar results shown by T98G cells ChIP assay also revealed that Snail directly regulated the transcription of Sox2 gene (Fig. 6D). As KMG4 cells possess more potent ability for inducing Snail-dependent Sox2 expression and forming colonies in soft agar comparing to T98G cells (Supplementary Fig. 5C and D), KMG4 cells were therefore utilized for subcutaneous injection in vivo, which revealed that 1 × 10² cells post irradiation had the potency to form tumors in SCID mice, whereas 1 × 10⁴ cells were required for the Snail-depleted cells (Table 1, Supplementary Fig. 6C).

Fig. 6.

Stemness gene expression was increased after 10 Gy irradiation in MG cells and abrogated by Snail knockdown. (A) Sox2, Oct3/4, and Nanog mRNA expression levels at 21 days after 10 Gy irradiation in control and Snail knockdown KMG4 cells were determined by semi-qRT-PCR. (B and C) Sox2 expression was determined by immunofluorescence analysis with an antibody to Sox2 (green), and F-actin was visualized by phalloidin (red). Relative Sox2 expression was graphed, as described in Materials and Methods. *P < .05. (D) Snail-overexpressing KMG4 and -T98G cells were subjected to ChIP assay, as described in Materials and Methods, and Snail-dependent promoter activity of Sox2 was determined.

Table 1.

Limiting dilution analysis for tumor incidence in NOD/SCID mice after subcutaneously injecting 10 Gy-irradiated KMG4 cells

Treatment	104	103	102
si Control	2/3
si Snail	1/3
si Control		2/3
si Snail		0/3
si Control			3/3
si Snail			0/3

Irradiation Induces Phosphorylation of GSK-3β and ERK in Snail-dependent Manner

Since the induction and activation of TGF-β1 have already been reported in the early phase after irradiation, we next investigated the concentration of TGF-β1 in MG cells in the status of pre- and postirradiation at 21 days. The lack of significant alteration in mRNA levels of TGF-β1 (Fig. 7A) and its production in the conditioned media (Fig. 7B), in addition to the phosphorylation levels of Smad2 (Fig. 7C), was shown in both T98G and KMG4 between pre- and postirradiated samples. Co-culture of glioma cells with either murine macrophages (RAW 264.7) or fibroblasts (NIH3T3) increased the expression levels of murine TGFβ mRNA post-irradiation suggesting a role for the microenvironment (FIg. 7D). In fact, Smad2/3 was phosphorylated upon stimulation of 5 ng/mL TGF-β1, which led to increases in Snail, MMP2, and MMP9 in both glioma cell lines (Fig. 7E and F). Phosphorylation of GSK-3β and ERK1/2, but not STAT3 and AKT, was remarkably increased at 21 days after irradiation, the effects of which were abolished by Snail knockdown (Fig. 8A). Treatment with MEK inhibitor (U0126) in T98G cells decreased the expression of Snail after irradiation in addition to diminishing the phosphorylation of ERK (Fig. 8B). Treatment with GSK-3β inhibitor (SB-216763) increased phosphorylation of GSK-3β (S6) followed by elevation of Snail after irradiation (Fig. 8C).

Fig. 7.

TGF-β1 mRNA levels are increased not in glioma cells but rather in fibroblasts and macrophages after irradiation. (A–C) T98G and KMG4 cells were treated with 10 Gy irradiation. At 21 days after irradiation, levels of human TGF-β1 mRNA (A) and the protein concentrations in conditioned media (B) were determined by qRT-PCR and ELISA, respectively. The data represent mean and SD of independent experiments performed in triplicate. (N.S., not significant). The cell lysates were subjected to immunoblotting, and the phosphorylation and expression levels of Smad2 protein were immunoblotted (C). (D) T98G and KMG4 cells were co-cultured with fibroblast NIH3T3 or macrophages RAW 264.7 derived from mouse, followed by irradiation, and mRNA levels of murine TGF-β1 were determined by qRT-PCR at 48 hours after irradiation. *P < .05 (E and F) KMG4 and T98G cells were stimulated with recombinant TGF-β1 at a concentration of 5 ng/mL and subjected to immunoblotting for phosphorylated Smad2/3 (E) and semi-qRT-PCR to determine mRNA levels of Snail, MMP-2, and MMP-9 (F). Abbreviation: N.S., not significant.

Fig. 8.

Irradiation induces phosphorylation of ERK1/2 and GSK-3β in a Snail-dependent manner. (A) T98G cells were transfected with Snail-targeting siRNA (#1; si1, #2; si2) or a control siRNA (siC), followed by 10 Gy irradiation. Immunoblotting was performed with antibodies to the indicated proteins. (B–C) T98G cells were pretreated with MEK inhibitor U0126 (B) or GSK-3β inhibitor SB216763 (C), followed by irradiation or no irradiation, respectively. Immunoblotting was performed 48 hours after irradiation with antibodies to the indicated proteins. (D) Hypothetical schema in this study is as follows: Irradiation induces Snail expression by the phosphorylations of ERK1/2 and GSK-3β, together with microenvironment-derived TGF-β-mediated Smad 2/3 phosphorylation in malignant glioma. Irradiation-induced Snail elicits both GMT and stemness-like phenotypes, and the depletion of Snail abrogated all of these phenotypes in addition to the phosphorylation of ERK1/2 and GSK-3β. In conclusion, irradiation induces Snail through a possible ERK1/2, GSK-3β,and TGF-β pathway, leading to GMT and stemness phenotypes in malignant glioma.

Discussion

Here we clarified that irradiation induced mesenchymal phenotypes in MG and that long-term elevation of Snail may contribute to transition to a mesenchymal phenotype of MG possibly via ERK and GSK-3β activation, together with TGF-β-dependent signaling pathway.

Currently, molecular classification for MG has been well established based on TCGA, and it has been reported that recurrent MG exhibited TCGA mesenchymal subtype in general. Consistent with this understanding, we observed the elevation of EMT-related mesenchymal markers such as vimentin, CD44, and YKL-40, which suggested TCGA-based mesenchymal subtype in postirradiated recurrent MG specimens and irradiated MG cell lines and primary cells.

To address the regulatory mechanism for postirradiated acquisition of mesenchymal features of MG, we focused on EMT regulators such as Slug, Snail, and Twist because they have been shown to play an important role in migration and invasion in MGs.^23,36 However, the EMT claims “canonical E-cadherin to N-cadherin switching” is unlikely to correlate with mesenchymal transition in glioma cells. Because glial cells are developmentally derived from the neuroepithelial lineage of cells, we employed the term “glial to mesenchymal transition (GMT)” instead of “EMT”.

In this study, single-fraction irradiation (10 Gy) and multifraction irradiation (2 Gy × 10 times) were used for irradiation. Both doses of irradiation dramatically induced apoptosis during the initial 2 weeks, and thereafter the remaining small population of cells started to re-proliferate (data not shown). We considered the expression profiles at 48 hours and 21 days after irradiation as reflecting the early and late events, respectively, and the properties of cells at 21 days after irradiation as more closely mimicking the resistance to clinical radiation. The expression of Snail was sustained at high levels in late phases, and the preceding knockdown of Snail before irradiation successfully abrogated the mesenchymal phenotypes such as invasion and motility. In contrast, knockdown of Slug in T98G cells had no effect on the expression of mesenchymal molecules (data not shown); thus Snail, rather than Slug, might be critical for regulating mesenchymal transition in MGs.

Furthermore, the expressions of both Slug and Twist mRNA were decreased in Snail knockdown T98G cells and increased in Snail-overexpressing cells (Supplementary Fig. 3A-C). In Snail knockdown cells, comprehensive decrease of other transcriptional factors, including Slug and Twist, may produce a substantial and broad effect on the expression of mesenchymal molecules. The expression of Twist seems to be controlled by Slug expression.³⁷

In general, irradiation-induced EMT has been shown to be regulated by multiple signalling pathways that can be divided into 2 groups based on TGF-β-dependency.^38,39 In the TGF-β-independent pathway, irradiation generates reactive oxygen species and activates ERK1/2,^40,41 leading to the nuclear localization of Snail.^42,43 Meanwhile, GSK-3β phosphorylates Snail, which leads to its degradation,⁴⁴ whereas GSK-3β per se has been reported to be inactivated by pERK1/2.⁴⁵

In this study, downregulation of Snail by siRNA was found to inhibit the irradiation-induced phosphorylation of ERK1/2 and GSK-3β, suggesting the presence of Snail-mediated activation of ERK1/2 and GSK-3β. As reported, we clarified that ERK1/2 and GSK-3β positively regulated Snail expression by inhibitor experiments (Fig. 8B and C). From these observations, we hypothesized that ERK/GSK-3β/Snail might constitute an activation loop and contribute to mediating the mesenchymal transition in MG after irradiation. To date, several researchers have demonstrated Snail-dependent ERK phosphorylation, but the underlying regulatory mechanism is still unknown.^46–48 Since irradiation did not increase the expression and phosphorylation of EGFR (Supplementary Fig. 7), we have not yet obtained any direct evidence of Snail-dependent ERK phosphorylation. Snail might regulate ERK phosphorylation through another growth factor and/or cytokine-dependent signaling pathway at the transcriptional level. Further study is necessary to address this issue in the future.

Considering the involvement of the TGF-β-dependent signaling mechanism in irradiation-induced mesenchymal transition, we examined the TGF-β1 production in MG cells. No significant increase in TGF-β1 concentration in culture media was observed after irradiation, nor was there a significant increase of the phosphorylation levels of Smad2 (Fig. 7A–C).

To investigate the role of tumor environment, we examined the effects of irradiation on the production of TGF-β in fibroblasts and macrophages co-cultured with MG cells. We found that the expression levels of TGF-β mRNA were increased in fibroblasts and macrophages after irradiation. We also confirmed that MG cells possess an intact signaling pathway to TGF-β because TGF-β stimulation phosphorylated Smad2/3 following the increased expression of Snail, MMP-2, and MMP9 in both T98G and KMG4 cells (Fig. 7D–F).

Thus, we conclude that mesenchymal transition in malignant glioma after irradiation is mediated by at least 2 pathways: a TGF-β-dependent effect derived from the surrounding tumor microenvironment⁴⁹ and a TGF-β-independent effect through ERK1/2 and GSK-3β. The previous studies described that TGF-β receptor inhibitor (LY2109761) enhanced radiation response and prolonged survival in GBM cells, and this may be due to inhibition of GMT of MG cells.^28,50

Recent reports also indicate the emergence of cancer stem cells (CSCs), in part with EMT features.^51,52 In our study, the increase in the stemness gene, such as Sox2, Oct3/4, and Nanog, was detected at 21 days after irradiation (Fig. 6A); Snail knockdown completely abrogated their expressions. High proliferative ability in surviving cells after irradiation and the inhibition by Snail knockdown was shown by the in vivo xenograft model (Table 1). Taken together with the in vitro and in vivo evidence, Snail may regulate stemness features in surviving cells after irradiation. Two possibilities can be considered to be the mechanism for the increase in stemness gene expression in irradiated glioma cells: one is the expansion of the irradiation-resistant glioma stem cell population, and the other is the reversion of differentiated MG cells to stem cells after irradiation. However, the precise mechanism underlying irradiation-induced expression of stemness genes especially mediated by Snail still remains obscure.⁵³ Several previous papers demonstrated that GBM stem cells (GSCs) are more resistant to radiation than matched non-stem glioma cells because GSCs preferentially enhance DNA damage checkpoint response through the activation of several checkpoint proteins such as ATM, Rad17, Chk2, and Chk1. In addition, GSCs are more efficient in repairing the damaged DNA and recover more rapidly from the genotoxic stress than matched non-stem tumor cells.⁵⁴ Thus, GSCs might display more resistance to radiation-induced apoptosis.

In summary, we emphasize that glioma cells post irradiation undergo GMT and stemness features throughout the early to late phases by upregulating Snail through the mechanisms of irradiation-induced activation of ERK1/2 and the subsequent inactivation of GSK3β, in addition to an activation of Smad2/3 evoked by TGFβ derived from tumor microenvironment. Ultimately, Snail seems to be a key factor for inducing both mesenchymal transition and stemness features in glioma cells after irradiation (Fig. 8D).

The mesenchymal phenotypes, along with their invasive characteristics and stemness features, are directly related to Snail expression levels. Concomitant inhibition of Snail with irradiation in malignant glioma may produce a greater effect by preventing GMT after irradiation and leading to the reduction in tumor aggressiveness in recurrence and the improvement of patients' prognoses.

Supplementary Material

Supplementary material is available online at Neuro-Oncology (http://neuro-oncology.oxfordjournals.org/).

Funding

Japan Society for the Promotion of Science (JSPS) Grant-in-Aid for Scientific Research (KAKENHI Grant Number 24590406) to M.T.