Necrosis is a hallmark of glioblastoma (GBM) and is responsible for poor prognosis and resistance to conventional therapies. However, the molecular mechanisms underlying necrotic microenvironment-induced malignancy of GBM have not been elucidated. Here, we report that transglutaminase 2 (TGM2) is upregulated in the perinecrotic region of GBM and triggered mesenchymal (MES) transdifferentiation of glioma stem cells (GSC) by regulating master transcription factors (TF), such as C/EBPβ, TAZ, and STAT3. TGM2 expression was induced by macrophages/microglia-derived cytokines via NF-κB activation and further degraded DNA damage–inducible transcript 3 (GADD153) to induce C/EBPβ expression, resulting in expression of the MES transcriptome. Downregulation of TGM2 decreased sphere-forming ability, tumor size, and radioresistance and survival in a xenograft mouse model through a loss of the MES signature. A TGM2-specific inhibitor GK921 blocked MES transdifferentiation and showed significant therapeutic efficacy in mouse models of GSC. Moreover, TGM2 expression was significantly increased in recurrent MES patients and inversely correlated with patient prognosis. Collectively, our results indicate that TGM2 is a key molecular switch of necrosis-induced MES transdifferentiation and an important therapeutic target for MES GBM. Cancer Res; 77(18); 4973–84. ©2017 AACR.

Glioblastoma multiforme (GBM) is the most malignant glial tumor and is associated with extremely poor survival (1). Current standard treatments, radiotherapy and chemotherapy have not improved the poor prognosis of GBM, which has a median overall survival of approximately 14 months and a 2-year survival rate of less than 10% (2). Analysis of large-scale gene expression and genomic datasets segregated GBM into four groups—proneural (PN), neural, classical (CL), and mesenchymal (MES; ref. 3). Notably, the MES subtype of GBM is associated with relatively poor prognosis compared with that of the other subtypes and shows resistance to conventional therapy (4). Moreover, several reports have suggested that MES transdifferentiation from other subtypes occurs during GBM progression due to the microenvironment or therapeutic stimuli (4–8). Therefore, it is very important to elucidate the detailed molecular mechanisms of MES transdifferentiation during GBM tumor progression to develop future therapeutics.

Necrosis, a hallmark of GBM, has been suggested to be responsible for the poor prognosis and resistance to conventional therapies (9, 10). The necrotic microenvironment contains infiltrating macrophages/microglia and glioma stem cells (GSC), and hypoxia-associated genes are expressed (11–14). Interestingly, patients with the MES subtype showed a high level of necrosis (3), and transcriptome analysis of the perinecrotic regions also showed high MES activity (15). Moreover, analysis of The Cancer Genome Atlas (TCGA) data set showed that non-MES GBMs became transcriptionally similar to the MES subtype with increasing levels of necrosis (15). Collectively, these results indicate that the necrotic microenvironment may be involved in MES transdifferentiation during GBM progression. A recent analysis of longitudinal genomic and transcriptomic data showed that expression-based subtype change is a major event of GBM recurrence (16). Because subtype transdifferentiation is affected by therapy-induced necrosis, we hypothesized that therapy-induced MES subtype change due to the necrotic microenvironment is the major cause of acquired resistance to conventional therapy and recurrence.

To test our hypothesis, we identified the molecular mechanisms underlying MES transdifferentiation in the necrotic regions of GBM. We analyzed common population of genes between necrosis-associated genes and MES GSC-enriched genes. Here, we showed that perinecrotic areas highly expressed transglutaminase 2 (TGM2), which contributed to the MES transdifferentiation of GSC via regulation of master transcription factors (TF) of MES GBM.

Cell culture

Astrocyte and 293T cells were maintained in DMEM supplemented with 10% FBS (HyClone). Human NSCs were purchased from Millipore and cultured according to the manufacturer's instructions. Patient-derived GBM stem cells (83NS, 131, 528NS, 84NS, 047T, and 352T2) were maintained in DMEM/F-12 supplemented with B27 (Invitrogen), EGF (10 ng/mL; R&D Systems), and bFGF (5 ng/mL; R&D Systems). Astrocyte was obtained from ScienCell Research Laboratories. 293T was obtained from ATCC. 83NS, 528NS, and 84NS were obtained from Dr. Ichiro Nakano (University of Alabama at Birmingham, Birmingham, AL). 131, 047T, and 352T2 were obtained from Dr. Do-Hyun Nam (Samsung Medical Center, Korea). All experiments using GSC were performed on cells in passages 5-30 (2015-2016). The cell cultures have not been authenticated but all cells were repeatedly screened for mycoplasma (e-Myco Mycoplasma PCR Detection Kit, iNtRON) and maintained in culture for less than 6 months after receipt.

Plasmids, transfection, and lentivirus production and infection

Lentivirus production was performed as previously reported (17). Briefly, 3 to 4 × 106 293T cells were plated on 100 mm culture dish 24 hours before transfection. Then, 4.5 μg of lentiviral construct (pHRST-IRES-TGM2, pLL3.7-shTGM2-B, and pLL3.7-shTGM2-C), 3 μg of psPAX2 (Addgene), and 1.5 μg of pMD2.G (Addgene) were cotransfected into 293T cells using 27 μL of Lipofectamine 2000 (Invitrogen). The medium was changed 6 hours after transfection. The medium containing lentivirus was harvested 48 hours after transfection. Viral particles were concentrated and purified using a Lenti-X concentrator (Clontech). Cells were infected with lentivirus in the presence of 6 μg/mL polybrene. siRNAs against human C/EBPβ, GADD153 and negative control siRNA (Genolution) were transfected in GSC using Lipofectamine 2000 (Invitrogen). The nucleotide sequences used for target-specific siRNA or shRNA are as follows: anti-GADD153 siRNA, 5′-GAAAGAACAGGAGAAUGAAUU-3′; shTGM2-B, 5′-GCCTCGTGGTTATTAGCAAGG-3′; and shTGM2-C, 5′-GCCATTGACCACCCACCATAT-3′.

Quantitative RT-PCR

Semiquantitative RT-PCR was performed as previously described (17, 18). Total RNA was isolated from cells using Trizol reagent (Invitrogen) according to the manufacturer's instructions. Total RNA (1 μg) was used as a template to synthesize cDNA using M-MLV reverse transcriptase (Invitrogen). The PCR primers are shown as follows: TGM2, sense 5′-AACATGGGCAGTGACTTTGA-3′ and antisense 5′-AGAGAAAGGCTCCAGGTTGA-3′; C/EBPβ, sense 5′-GACAAGCACAGCGACGAGTA-3′ and antisense 5′-CAGCTGCTCCACCTTCTTCT-3′; TNFα, sense 5′-ACGGCATGGATCTCAAAGAC-3′ and antisense 5′-GTGGGTGAGGAGCACGTAGT-3′; GADD153, sense 5′-AGATGGCAGCTGAGTCATTG-3′ and antisense 5′-GTTCTGGCTCCTCCTCAGTC-3′; and GAPDH, sense 5′-GGAGTCCACTGGCGTCTTCAC-3′ and antisense 5′-GAGGCATTGCTGATGATCTTGAGG-3′. The PCR products were analyzed on a 1% agarose gel.

Limiting dilution assay

For in vitro limiting dilution assays (LDA), GSC with decreasing numbers of cells (200, 100, 50, and 10 or 100, 50, 25, and 5) per well were plated in 96-well plates containing DMEM/F-12 with B27, EGF (10 ng/mL), and bFGF (5 ng/mL). Extreme limiting dilution analysis was performed using software available at http://bioinf.wehi.edu.au/software/elda/.

Immunoblot analysis

Proteins were extracted with RIPA buffer with complete protease inhibitors (Roche), separated by electrophoresis, transferred to PVDF membranes (Millipore), and blocked with 5% skim milk (BD Biosciences). Primary antibodies against TGM2 (Chemicon), CD44 (R&D Systems), Sox2 (R&D Systems), TAZ (Cell Signaling Technology), C/EBPβ (Santa Cruz Biotechnology), p-STAT3 (Cell Signaling Technology), STAT3 (Santa Cruz Biotechnology), IκBα (Santa Cruz Biotechnology), GADD153 (Santa Cruz Biotechnology), and α-tubulin (Santa Cruz Biotechnology) were incubated overnight at 4°C. Immunoreactive bands were visualized using peroxidase-labeled affinity purified secondary antibodies (KPL) and the Amersham ECL prime Western blotting detection reagent (GE Healthcare).

Polymerization of GADD153 by TGM2

Purified guinea pig liver transglutaminase (Zedira) was diluted in buffer containing 10 mmol/L Tris-HCl (pH 8.0), 1 mmol/L EDTA (pH 8.0), and 0.1 mol/L NaCl. The TGM2 2 was incubated in 40 μL of reaction buffer [50 mmol/L Tris-HCl (pH 8.0), 50 mmol/L NaCl, 0.5 mmol/L EDTA, 1 mmol/L MgCl2, 5 mmol/L DTT, 10 mmol/L CaCl2] with human GADD153 (Biosource) at 37°C for 10 minutes. The reaction mixture was separated using 4% to 12% Tris-glycine gels (Invitrogen).

In vivo study

All animal experiments were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee at the National Cancer Center, Republic of Korea. For the orthotopic mouse model (19, 20), cells were transplanted following resuspension in DMEM/F-12 with B27, EGF (10 ng/mL), and bFGF (5 ng/mL). Cells were injected stereotactically into the left striatum of 5-week-old female BALB/c nude mice. The injection coordinates were 2.2 mm to the left of the midline and 0.2 mm posterior to the bregma at a depth of 3.5 mm. The brain of each mouse was harvested and fixed in 4% paraformaldehyde. For the subcutaneous mouse model, cells were injected into the hip area on both sides of each mouse. Tumor growth was measured two times a week using electronic caliper to measure two diameters using the following formula: length × width2 × 0.5. The mean tumor volume at the start of the experiment was approximately 200 mm3. Mice were sacrificed 4 weeks after the cell injections. The tumors were extracted, pooled for each experimental group, and mechanically disaggregated using steel operating scissors. GK921 (2 mg/kg; MedChem Express) was orally administered daily and radiation treatment was administered at a dose of 2.5 Gy for 4 days. Survival and tumor size were analyzed using GraphPad PRISM software (version 7).

Histology and IHC staining

For observation of the histologic features, the brains were removed, fixed with 4% paraformaldehyde for 24 hours at 4°C, and stained with hematoxylin (DaKo) and 0.25% eosin (Merck). For IHC staining of cancer stem cell markers and TGM2-associated genes [CD44 (Cell Signaling Technology), Sox2 (R&D Systems), TGM2 (Cell Signaling Technology), C/EBPβ (Santa Cruz Biotechnology), GADD153 (Santa Cruz Biotechnology)], after the antigen retrieval process with citrate buffer (pH 6.0) and endogenous peroxidase blocking with 3% hydrogen peroxide, tissue sections were incubated in primary antibody overnight at 4°C in a humidified chamber with IHC world antibody diluent buffer. We developed samples using 3,3′-diaminobenzidine (DAB, Vector Laboratories) as the chromogen. To quantify the IHC staining, immunostained cells were analyzed using a Tissue FAXS system (Tissue Gnostics). The scanned images were quantified using the HistoQuest cytometry software in the Tissue FAXS system. The cut-off threshold was determined by the stained density of the negative control. For patient tissue staining, GBM specimens were obtained from patients undergoing surgery with the approval of the Institutional Review Board of the Samsung Medical Center (No. 2010-04-004) and with informed consent. This Institutional Review Board follows the guidelines of Declaration of Helsinki and Belmont Report.

Bioinformatics analysis

Genomic data and clinicopathologic data for GBM samples were downloaded from TCGA data portal (http://cancergenome.nih.gov/). For TCGA RNA sequencing data analysis, BAM files were available from the Cancer Genomics Hub (http:/cghub.ucsc.edu). The expression measurements, RPKM values were estimated using R package DEGseq. The nearest template prediction algorithm was used to predict the MES activity of a given TCGA samples with statistical significance using a predefined set of markers that are specific to MES subtype (21, 22). We also analyzed clinical information for the patients with untreated GBM who had survival information available.

As shown in Fig. 1A, to reanalyze the molecular characteristics of cancer stem cells in our study, we acquired microarray data in raw form from the Gene Expression Omnibus (GEO) database (GEO number, GSE67089), published at Ichiro Nagano Laboratory (7). Affymetrix Human U219 CEL files (GSE67089) were normalized using the robust multiarray average method from the Affy packages (23). For unbiased analysis, unspecific filtering was performed with the SD filter (24), and genes that were not changed or were expressed at a very low level at all microarrays were removed by the cut-off value (SD less than 2). The Hopkins statistic (25) was used to assess the clustering tendency of the filtered dataset. The filtered dataset was highly clusterable (H value, 0.1900499). The Elbow method (26) was used to determine the optimal number of the clusters and four clusters were suggested. All samples from the dataset were grouped into four subgroups (astrocytes, glioma cell lines, cancer stem cells of two subtypes, PN and MES) using the filtered genes (total 901 genes) through principle component analysis (PCA), K-means (27), PAM (28), and hierarchical clustering.

Figure 1.

TGM2 is highly expressed in the MES subtype of GBM. A, The overlapping genes correlated with necrosis (TCGA database) and the MES subtype (I. Nakano database). Details of the analysis are described in Materials and Methods. B, IHC analysis of TGM2 staining in GBM patient tissues. N, necrotic region; PN, perinecrotic region. Scale bar, 100 μm. C, TGM2 expression fold changes in non-MES and MES tissues were determined by RNA sequencing in the initial and recurrent samples. D, TGM2 expression in each GBM subtype from TCGA database. E, MES activity ratio of TGM2 high and low expression in TCGA database. F, GBM patient survival graphs comparing between TGM2 high expression group (high in 20% of the samples in TCGA database) and others.

Figure 1.

TGM2 is highly expressed in the MES subtype of GBM. A, The overlapping genes correlated with necrosis (TCGA database) and the MES subtype (I. Nakano database). Details of the analysis are described in Materials and Methods. B, IHC analysis of TGM2 staining in GBM patient tissues. N, necrotic region; PN, perinecrotic region. Scale bar, 100 μm. C, TGM2 expression fold changes in non-MES and MES tissues were determined by RNA sequencing in the initial and recurrent samples. D, TGM2 expression in each GBM subtype from TCGA database. E, MES activity ratio of TGM2 high and low expression in TCGA database. F, GBM patient survival graphs comparing between TGM2 high expression group (high in 20% of the samples in TCGA database) and others.

Close modal

A web tool for Venn diagrams (bioinformatics.psb.ugent.be/webtools/Venn/) was used to select common genes between the above 919 genes (unique gene, 561 genes) and 1,150 positive necrosis-related genes (unique gene, 864 genes). The tool identified 90 common genes and the variable factors and separated all samples into four groups using PCA. Through this analysis, the 90 genes were revealed to be important variables associated with each group (astrocytes, glioma cell lines, and PN and MES subtypes). Next, due to the association between the MES subtype and necrotic characteristics of GBM, significant analysis of microarray (SAM) was performed using the freely available MeV software (version 4.9.0, http://mev.tm4.org) to determine whether there were differentially expressed genes in MES subtype cancer stem cells compared to PN subtype. Twenty-seven genes were found to be statistically significant in the analysis, and the order of genes was ranked by fold change. This analysis was performed by R and MeV software in all bioinformatics analyses.

Statistical analysis

All the data are expressed as the mean ± SEM from at least three independent experiments. For survival curve analysis, the Kaplan–Meier method was used. In the case of patients who were alive at the time of last follow-up, the survival records were censored in our analysis. Statistical Package for the Social Sciences software, version 16 (SPSS) was used for statistical analysis. Spearman's correlation coefficients and significance (two-tailed) were calculated using SPSS software for each pair of genes. In the case of the mouse experiments, results of the multi-dataset were compared by analysis of variance using the log-rank (Mantel–Cox) test. The results of the two-dataset experiments were compared using a two-tailed Student t test. The level of statistical significance stated in the text is based on the P values. *, P < 0.05 or **, P < 0.01 was considered statistically significant.

TGM2 is upregulated in MES GBM

Given that patients with the MES subtype exhibit a high degree of tumor necrosis, we compared the gene signature between the tumor necrotic regions and MES GSC (3, 7). TGM2 was selected as a strong candidate because of its important role in tumorigenesis (Fig. 1A). By using tissue microarray (TMA) analysis, we confirmed that TGM2 was upregulated in MES GBM samples and enriched in the perinecrotic regions of human GBM tissues (Supplementary Fig. S1A–C and Fig. 1B). Furthermore, recurrent MES patient samples showed significant increases in TGM2 expression compared to initial MES samples (Fig. 1C). Our results raised possibility that TGM2 might play an important role for the MES transdifferentiation in recurrent MES GBM.

Among the members of the transglutaminase family, TGM2 is expressed in various cancer cell types and associated with poor patient survival (29–34). However, the functions of TGM2 in GBM, especially in the MES transdifferentiation of GSC, are unclear. To analyze clinical importance of TGM2 expression in GBM, we carried out an expression analysis from TCGA data set. TGM2 was highly expressed in the MES subtype (Fig. 1D) and showed strong MES activity in the MES samples (Fig. 1E). Furthermore, the prognosis for GBM patients with high TGM2 expression was poor (Fig. 1F).

Taken together, the results showed that TGM2 can be used as a prognostic marker for GBM, because of the markedly higher expression of TGM2 in GBM patients with the malignant MES subtype and a poor prognosis.

TGM2 regulates the transdifferentiation of MES GSC

To address functional significances of TGM2, we first confirmed TGM2 expression in subtypes of GSC. In immunoblot analysis, TGM2 was detected in the MES GSC but not in astrocytes, neural stem cells (NSC), or PN and CL GSC (Fig. 2A). The expression of CD44, a MES subtype marker, was similar to that of TGM2, whereas Sox2, a marker of the PN subtype, was inversely related to the expression of TGM2 (Fig. 2A; refs. 35–37). Similar patterns were also observed in an IHC analysis of a GSC-injected orthotopic mouse model (Fig. 2B).

Figure 2.

TGM2 regulates stemness of MES and PN GSC. A, Immunoblot (IB) and PCR analysis showing the concentration of TGM2, CD44, and Sox2 in GBM patient-derived cells. α-Tubulin and GAPDH were used as loading controls. B, IHC analysis of TGM2, CD44, and Sox2 in a xenograft mouse model established using GSC. C, Immunoblot analysis of TGM2, CD44, TAZ, C/EBPβ, pSTAT3, and STAT3 in GSC (83NS) infected with shTGM2-expressing lentiviral or control construct. α-Tubulin was used as a loading control. D, LDA was performed in 83NS-Con, 83NS-shTGM2-B, and 83NS-shTGM2-C cells. E, Immunoblot analysis of TGM2, CD44, TAZ, C/EBPβ, pSTAT3, and STAT3 in GSC (528NS) infected with TGM2-expressing lentiviral or control construct. α-Tubulin was used as a loading control. F, LDA was performed in 528NS-Con and 528NS-TGM2 cells.

Figure 2.

TGM2 regulates stemness of MES and PN GSC. A, Immunoblot (IB) and PCR analysis showing the concentration of TGM2, CD44, and Sox2 in GBM patient-derived cells. α-Tubulin and GAPDH were used as loading controls. B, IHC analysis of TGM2, CD44, and Sox2 in a xenograft mouse model established using GSC. C, Immunoblot analysis of TGM2, CD44, TAZ, C/EBPβ, pSTAT3, and STAT3 in GSC (83NS) infected with shTGM2-expressing lentiviral or control construct. α-Tubulin was used as a loading control. D, LDA was performed in 83NS-Con, 83NS-shTGM2-B, and 83NS-shTGM2-C cells. E, Immunoblot analysis of TGM2, CD44, TAZ, C/EBPβ, pSTAT3, and STAT3 in GSC (528NS) infected with TGM2-expressing lentiviral or control construct. α-Tubulin was used as a loading control. F, LDA was performed in 528NS-Con and 528NS-TGM2 cells.

Close modal

To determine the role of TGM2 in transdifferentiation of MES, we silenced TGM2 in MES GSC (83NS and 131) using shRNA, examined the levels of master TF, including C/EBPβ, TAZ, and STAT3, and performed LDAs. When TGM2 was silenced in MES GSC, both the MES marker CD44 and master TF decreased (Fig. 2C and Supplementary Fig. S2A). The sphere-forming ability was decreased when TGM2 was suppressed in MES subtype cells, as shown in LDAs measuring the sphere-forming ability of cancer stem cells (Fig. 2D and Supplementary Fig. S2B; ref. 38). Next, we overexpressed TGM2 in PN 528NS and 84NS GSC (7). The results showed that overexpression of TGM2 increased the MES marker CD44 and master TF (Fig. 2E and Supplementary Fig. S2C). Similarly, the GSC sphere-forming ability was enhanced as TGM2 expression increased (Fig. 2F and Supplementary Fig. S2D). We obtained similar results using 047T and 352T2 cells, CL subtype cells (Supplementary Fig. S2E–H). These results indicate that TGM2 regulate self-renewal activity of MES GSC by modulation of master TF, which is critical for maintaining the transcriptome profiling of the MES subtype.

TGM2 regulates tumor progression in vivo

To further analyze the functional significance of TGM2 in vivo, we generated an orthotopic mouse model using TGM2-silenced 83NS cells. Depletion of TGM2 in MES GSC significantly increased mouse survival and reduced tumor size compared to that of the control GSC (Fig. 3A and B). In addition, the expression levels of TGM2 and the MES marker CD44 were specifically decreased in 83NS cells with TGM2 knockdown as determined by IHC (Fig. 3B). Moreover, overexpression of TGM2 in PN GSC significantly decreased mouse survival and increased tumor size and the level of TGM2 and CD44 compared to those of the control GSC (Fig. 3C and D). Taken together, these results suggest TGM2 to be an important therapeutic target in MES GBM.

Figure 3.

TGM2 regulates tumorigenicity of MES and PN GSC in an orthotopic xenograft mouse model. A, Kaplan–Meier survival graph of mice implanted with 83NS-Con, 83NS-shTGM2-B, and 83NS-shTGM2-C cells (n = 7, 1 × 104 cells injected in each mouse). B, Hematoxylin and eosin (H&E) staining of the whole brain and IHC analysis of TGM2 and CD44 in the orthotopic xenograft mouse model. Scale bar, 100 μm. C, Kaplan–Meier survival graph of mice implanted with 528NS-Con and 528NS-TGM2 cells (n = 6, 5 × 104 cells injected in each mouse). D, Hematoxylin and eosin (H&E) staining of the whole brain and IHC analysis of TGM2 and CD44 in the orthotopic xenograft mouse model. Scale bar, 100 μm.

Figure 3.

TGM2 regulates tumorigenicity of MES and PN GSC in an orthotopic xenograft mouse model. A, Kaplan–Meier survival graph of mice implanted with 83NS-Con, 83NS-shTGM2-B, and 83NS-shTGM2-C cells (n = 7, 1 × 104 cells injected in each mouse). B, Hematoxylin and eosin (H&E) staining of the whole brain and IHC analysis of TGM2 and CD44 in the orthotopic xenograft mouse model. Scale bar, 100 μm. C, Kaplan–Meier survival graph of mice implanted with 528NS-Con and 528NS-TGM2 cells (n = 6, 5 × 104 cells injected in each mouse). D, Hematoxylin and eosin (H&E) staining of the whole brain and IHC analysis of TGM2 and CD44 in the orthotopic xenograft mouse model. Scale bar, 100 μm.

Close modal

TGM2 regulates C/EBPβ expression directly by polymerization of GADD153

To determine whether TGM2 is the key molecular switch of MES transdifferentiation, we first analyzed the correlation between TGM2 and the master TF using TCGA dataset. Because C/EBPβ was most strongly correlated with TGM2 expression among MES master TF (Fig. 4A and Supplementary Fig. S3A), we assessed the effect of TGM2 on the regulation of C/EBPβ. Previous reports demonstrated that TGM2 degrades target proteins by crosslinking; therefore, we speculated that there must be a negative upstream regulator of C/EBPβ as a TGM2 substrate, and this protein may be degraded by TGM2-mediated crosslinking (31, 32). GADD153, a nuclear protein that avidly dimerizes with C/EBP isoforms α and β, is a transcriptional inhibitor of C/EBPβ signaling (39, 40). Moreover, induction of GADD153 was also associated with decreased expression of C/EBPβ (40). As GADD153 expression showed anticorrelation with C/EBPβ in TCGA dataset (Fig. 4B), we hypothesized that TGM2-mediated GADD153 crosslinking may induce C/EBPβ expression. When TGM2 was overexpressed in 528NS cells, the expression of C/EBPβ increased, and GADD153 expression decreased (Fig. 4C). In contrast, suppressing TGM2 expression in 83NS cells led to an increase in GADD153 (Fig. 4D). More importantly, silencing of GADD153 in 528NS cells strongly induced C/EBPβ expression, suggesting that GADD153 is a negative regulator of C/EBPβ (Fig. 4E). To confirm the degradation of GADD153, we added proteasome inhibitors to MES subtype cells and found that the expression of GADD153 was augmented (Fig. 4F and Supplementary Fig. S3B and C). To further determine whether GADD153 is a direct substrate of TGM2, we incubated recombinant human GADD153 with purified TGM2 in vitro. TGM2 induced the polymerization of GADD153 in a Ca2+ -dependent manner (Fig. 4G and H).

Figure 4.

TGM2 regulates C/EBPβ expression directly by polymerization of GADD153. A, A correlation dot plot of C/EBPβ and TGM2 in TCGA database. B, A correlation dot plot of GADD153 and TGM2 in TCGA database. C, Immunoblot analysis of TGM2, GADD153, and C/EBPβ in 83NS-Con, 83NS-TGM2 cells. α-Tubulin was used as a loading control. D, Immunoblot analysis of TGM2, GADD153, and C/EBPβ in 83NS-Con, 83NS-shTGM2-B, and 83NS-shTGM2-C cells. α-Tubulin was used as a loading control. E, Immunoblot (IB) and PCR analysis of GADD153 and C/EBPβ in 528NS transfected with siC/EBPβ or siControl. α-Tubulin was used as a loading control. F, Immunoblot analysis of TGM2 and GADD153 in 83NS cells treated with the proteasome inhibitor MG132 (10 μmol/L, 12 hours) and epoxomicin (2 μmol/L, 12 hours). α-Tubulin was used as a loading control. G and H, Immunoblot analysis of TGM2 (G) and GADD153 (H) in vitro cultured with CaCl2 (10 mmol/L), TGM2 recombinant protein (0, 50, 250, and 500 ng), or GADD153 recombinant protein (500 ng).

Figure 4.

TGM2 regulates C/EBPβ expression directly by polymerization of GADD153. A, A correlation dot plot of C/EBPβ and TGM2 in TCGA database. B, A correlation dot plot of GADD153 and TGM2 in TCGA database. C, Immunoblot analysis of TGM2, GADD153, and C/EBPβ in 83NS-Con, 83NS-TGM2 cells. α-Tubulin was used as a loading control. D, Immunoblot analysis of TGM2, GADD153, and C/EBPβ in 83NS-Con, 83NS-shTGM2-B, and 83NS-shTGM2-C cells. α-Tubulin was used as a loading control. E, Immunoblot (IB) and PCR analysis of GADD153 and C/EBPβ in 528NS transfected with siC/EBPβ or siControl. α-Tubulin was used as a loading control. F, Immunoblot analysis of TGM2 and GADD153 in 83NS cells treated with the proteasome inhibitor MG132 (10 μmol/L, 12 hours) and epoxomicin (2 μmol/L, 12 hours). α-Tubulin was used as a loading control. G and H, Immunoblot analysis of TGM2 (G) and GADD153 (H) in vitro cultured with CaCl2 (10 mmol/L), TGM2 recombinant protein (0, 50, 250, and 500 ng), or GADD153 recombinant protein (500 ng).

Close modal

These collective data indicate that increased expression of TGM2 in MES subtype cells leads to the direct degradation of GADD153 and thus induces C/EBPβ expression.

Radiation-induced TGM2 regulates MES transdifferentiation

Among the current therapeutic interventions for GBM, radiotherapy often leads to recurrence, largely due to the survival of radioresistant GSC (41, 42). Following radiotherapy, GSC shift from the PN to the MES subtype, mainly due to the secretion of TNFα by infiltrating macrophages/microglia (4). However, the molecular link between macrophages/microglia-secreted cytokines and MES transdifferentiation has not been clearly demonstrated. To determine the possible role of TGM2 in radiation-induced MES transdifferentiation by macrophages/microglia infiltration, we analyzed macrophages/microglia infiltration and TGM2 expression in 83NS-derived xenograft tumors with or without radiation treatment. As expected, radiation treatment induced TGM2 expression and macrophages/microglia infiltration (IBA1 staining) in the perinecrotic regions (Fig. 5A). Furthermore, in response to irradiation treatment on 528NS cells, the expression of TGM2 was increased and, consequently, the expression of master TF (Fig. 5B) and sphere forming ability was also enhanced (Fig. 5C). More importantly, we confirmed that radiation induced TGM2 and TNFα expression in 528NS-derived xenograft tumor (Fig. 5D). When the recombinant TNFα was applied, the expression of TGM2, CD44, and master TF increased in a dose-dependent manner (Fig. 5E). In contrast, expression of Iκ-Bα, a negative regulator of NF-κB, decreased (Fig. 5E). To demonstrate the molecular link between NF-κB and this event, we treated Bay 11-7085, an irreversible inhibitor of Iκ-Bα phosphorylation, to PN GSC and found that the decreased expression of Iκ-Bα by TNFα treatment was rescued, and TGM2 expression was downregulated (Fig. 5F). Moreover, downstream targets of TGM2, CD44, and master TF were also downregulated (Fig. 5F). Furthermore, we investigated whether master TF were regulated by TGM2 following NF-κB activation. As shown in Fig. 5G, TNFα induced master TF expression was blocked by TGM2 silencing. Taken together, the results indicate that after irradiation of PN GSC, TGM2 is upregulated via NFκB signaling to enhance transdifferentiation to MES subtype and acquire radioresistance.

Figure 5.

Radiation-induced TGM2 regulates MES transdifferentiation. A, Immunocytochemistry analysis of TGM2 and IBA1 expression in GBM xenografts derived from MES 83NS with or without radiation treatment. TGM2 was labeled in green and IBA1 was labeled in red. Nuclei were counterstained with DAPI (blue). N, necrotic region. Scale bar, 100 μm. B, Immunoblot analysis of TGM2, CD44, TAZ, C/EBPβ, pSTAT3, and STAT3 in 528NS cells, which were treated with radiation (3 Gy). α-Tubulin was used as a loading control. C, LDA was performed in 528NS-Con and 528NS-IR cells. D, PCR analysis showing the RNA concentration of TNFα and TGM2 in 528NS-derived xenograft tumor. GAPDH were used as loading controls. E, Immunoblot analysis of IκBα, TGM2, CD44, TZA, C/EBPβ, pSTAT3, and STAT3 in 528NS cells treated with TNFα recombinant protein (0, 10, 20, and 50 ng/mL). α-Tubulin was used as a loading control. F, Immunoblot analysis of IκBα, TGM2, CD44, C/EBPβ, pSTAT3, and STAT3 in 528NS cells treated with TNFα recombinant protein (100 μmol/L) and an IκBα inhibitor (Bay 11-7085, 10 μmol/L). α-Tubulin was used as a loading control. G, Immunoblot analysis of TGM2, CD44, C/EBPβ, pSTAT3, and STAT3 in 528NS cells infected with shTGM2-expressing lentiviral or control construct treated with TNFα recombinant protein (100 μmol/L). α-Tubulin was used as a loading control.

Figure 5.

Radiation-induced TGM2 regulates MES transdifferentiation. A, Immunocytochemistry analysis of TGM2 and IBA1 expression in GBM xenografts derived from MES 83NS with or without radiation treatment. TGM2 was labeled in green and IBA1 was labeled in red. Nuclei were counterstained with DAPI (blue). N, necrotic region. Scale bar, 100 μm. B, Immunoblot analysis of TGM2, CD44, TAZ, C/EBPβ, pSTAT3, and STAT3 in 528NS cells, which were treated with radiation (3 Gy). α-Tubulin was used as a loading control. C, LDA was performed in 528NS-Con and 528NS-IR cells. D, PCR analysis showing the RNA concentration of TNFα and TGM2 in 528NS-derived xenograft tumor. GAPDH were used as loading controls. E, Immunoblot analysis of IκBα, TGM2, CD44, TZA, C/EBPβ, pSTAT3, and STAT3 in 528NS cells treated with TNFα recombinant protein (0, 10, 20, and 50 ng/mL). α-Tubulin was used as a loading control. F, Immunoblot analysis of IκBα, TGM2, CD44, C/EBPβ, pSTAT3, and STAT3 in 528NS cells treated with TNFα recombinant protein (100 μmol/L) and an IκBα inhibitor (Bay 11-7085, 10 μmol/L). α-Tubulin was used as a loading control. G, Immunoblot analysis of TGM2, CD44, C/EBPβ, pSTAT3, and STAT3 in 528NS cells infected with shTGM2-expressing lentiviral or control construct treated with TNFα recombinant protein (100 μmol/L). α-Tubulin was used as a loading control.

Close modal

Knockdown of TGM2 suppresses radiation-induced MES transdifferentiation in vivo

To interrogate the functional relations of TGM2 in radio-resistance, we confirmed the combined effect of TGM2 knockdown and radiation treatment in the orthotopic mouse model. A combination of TGM2 silencing and radiation treatment further prolonged mouse survival, and decreased in tumor size of an 83NS-derived model (Fig. 6A and B). Notably, IHC analysis showed that TGM2, CD44, and IBA1 expression further increased after radiation treatment and blocked by TGM2 silencing (Fig. 6B). These results suggest that TGM2 induction by radiation may play a critical role in radioresistance of MES GSC.

Figure 6.

Suppression of TGM2 inhibits radiation-induced MES transdifferentiation in an orthotopic xenograft mouse model. A, Kaplan–Meier survival graph of mice implanted with 83NS cells infected with shTGM2-expressing lentiviral or control construct with combinational treatment of radiation (n = 4, 1 × 104 cells injected in each mouse). B, Hematoxylin and eosin (H&E) staining of the whole brain and IHC analysis of TGM2, CD44, and IBA1 in the orthotopic xenograft mouse model of A. Scale bar, 100 μm. C, Kaplan–Meier survival graph of mice implanted with 528NS cells infected with shTGM2-expressing lentiviral or control construct and treated with radiation (n = 6, 1 × 104 cells injected in each mouse). D, Hematoxylin and eosin staining of the whole brain and IHC analysis of TGM2 in the orthotopic xenograft mouse model of C. Scale bar, 100 μm.

Figure 6.

Suppression of TGM2 inhibits radiation-induced MES transdifferentiation in an orthotopic xenograft mouse model. A, Kaplan–Meier survival graph of mice implanted with 83NS cells infected with shTGM2-expressing lentiviral or control construct with combinational treatment of radiation (n = 4, 1 × 104 cells injected in each mouse). B, Hematoxylin and eosin (H&E) staining of the whole brain and IHC analysis of TGM2, CD44, and IBA1 in the orthotopic xenograft mouse model of A. Scale bar, 100 μm. C, Kaplan–Meier survival graph of mice implanted with 528NS cells infected with shTGM2-expressing lentiviral or control construct and treated with radiation (n = 6, 1 × 104 cells injected in each mouse). D, Hematoxylin and eosin staining of the whole brain and IHC analysis of TGM2 in the orthotopic xenograft mouse model of C. Scale bar, 100 μm.

Close modal

In previous results, we demonstrated that PN to MES transition was mediated by TGM2 expression; thus, we postulated that TGM2 expression may increase during irradiation of PN GSC to induce recurrence and radioresistance via MES subtype change. To verify this hypothesis, we measured the expression of TGM2 in a 528NS animal model after irradiation. As expected, TGM2 expression was increased in the irradiated tissues of the 528NS-injected mouse model (Fig. 6C and D). These results suggest that PN subtype patients may express TGM2 during irradiation to promote the MES subtype change. Thus, inhibition of TGM2 to block PN to MES transition could be a good therapeutic strategy. To assess this hypothesis, we analyzed the combination treatment of irradiation and TGM2 silencing using the 528NS animal model; survival of the 528NS-injected mice was prolonged in the radiation treatment group, but this group eventually showed radioresistance due to TGM2 induction, as expected (Fig. 6C and D). TGM2 silencing in the radiation combined group substantially increased survival of the 528NS-injected mice (Fig. 6C). Taken together, our results strongly suggest that a combination of TGM2-specific suppression and conventional radiotherapy may be effective not only in MES GBM but also non-MES GBM patients.

The TGM2 inhibitor GK921 specifically induces cell death and reduces the MES signature and tumorigenesis of MES GSC

To explore possible clinical application, we evaluated the effects of blocking TGM2 using GK921, a TGM2-specific inhibitor (43, 44). When GK921 was applied, cell growth of the astrocytes and PN subtype cells was not significantly affected, whereas in MES subtype cells, cell viability was reduced even at a low concentration of 0.5 μmol/L (Fig. 7A). In addition, CD44 and master TF were reduced following treatment with GK921 (Fig. 7B and C). To further confirm in vivo efficacy of its treatment, we generated an orthotopic xenograft mouse model using MES GSC (83NS) and orally administered GK921. The results showed that the median survival increased, and the tumor size was also reduced by GK921 treatment (Fig. 7D and E). Notably, IHC analysis demonstrated that CD44 and C/EBPβ expression was reduced, and GADD153 expression increased in the GK921 treatment group (Fig. 7E). Because we suggested significant therapeutic benefit of radiation therapy and TGM2 inhibition in non-MES GSC by blocking non-MES to MES subtype change, we tested GK921 drug effect combined with irradiation in 528NS-derived subcutaneous mouse model. Although GK921 or irradiation treatment partially reduced tumor volume compared with vehicle group, but these groups showed relapse of tumor growth (Fig. 7F). However, combined treatment of GK921 and irradiation significantly decreased tumor size and prolonged duration of treatment response (Fig. 7F).

Figure 7.

GK921 inhibits MES GSC growth and stemness. A, Cell viability assays of astrocytes and GBM patient-derived cells treated with different concentrations of GK921 (48 hours). B and C, Immunoblot analysis of TGM2, CD44, TAZ, C/EBPβ, pSTAT3, and STAT3 in MES GSC [83NS (B) and 131 (C)] treated with GK921 (24 hours). D, Kaplan–Meier survival graph of mice implanted with 83NS cells treated with GK921 (n = 12, 5 × 102 cells injected in each mouse). E, Hematoxylin and eosin (H&E) staining of the whole brain and IHC analysis of CD44, C/EBPβ, and GADD153 in a GK921-treated orthotopic xenograft mouse model. Scale bar, 100 μm. F, A graph comparing tumor volume in a vehicle, GK921 treatment, radiation treatment, and the combinational treatment at subcutaneous xenograft mouse model [vehicle (n = 4), GK921 (n = 4), IR (n = 3), GK921+IR (n = 4), 1 × 106 cells injected in each mouse].

Figure 7.

GK921 inhibits MES GSC growth and stemness. A, Cell viability assays of astrocytes and GBM patient-derived cells treated with different concentrations of GK921 (48 hours). B and C, Immunoblot analysis of TGM2, CD44, TAZ, C/EBPβ, pSTAT3, and STAT3 in MES GSC [83NS (B) and 131 (C)] treated with GK921 (24 hours). D, Kaplan–Meier survival graph of mice implanted with 83NS cells treated with GK921 (n = 12, 5 × 102 cells injected in each mouse). E, Hematoxylin and eosin (H&E) staining of the whole brain and IHC analysis of CD44, C/EBPβ, and GADD153 in a GK921-treated orthotopic xenograft mouse model. Scale bar, 100 μm. F, A graph comparing tumor volume in a vehicle, GK921 treatment, radiation treatment, and the combinational treatment at subcutaneous xenograft mouse model [vehicle (n = 4), GK921 (n = 4), IR (n = 3), GK921+IR (n = 4), 1 × 106 cells injected in each mouse].

Close modal

On the basis of these results, we concluded that GK921 increased response to radiation therapy of GSC by blocking MES transdifferentiation. Our results strongly support the therapeutic potential of TGM2 inhibition in GBM patients.

Subtype changes of GBM are strongly associated with acquisition of resistance to conventional therapy and poor prognosis. Because MES transdifferentiation leads to recurrence and resistance to conventional therapies, the underlying mechanisms of this shift require elucidation for therapeutic improvement (4, 7). Our study demonstrated TGM2 as a key molecular switch for the necrosis-induced MES transdifferentiation. TGM2 modulates this process by inducing proteasomal degradation through the polymerization of GADD153 and by modulating C/EBPβ. Silencing and pharmacologic inhibition of TGM2 significantly increased survival rate and therapeutic efficacy of irradiation in mouse models. In addition to extensive in vivo data, clinical relevance of our findings was supported by in silico analysis of TGM2 expression in a large glioma genomic dataset and IHC analysis of patient-derived specimens. Especially, TGM2 expression was highly upregulated in recurrent MES patient samples compared to that of non-MES and initial GBM samples, and correlated with poor prognosis. These results suggested that TGM2 might play an important role for the recurrence and radioresistance of GBM by inducing MES transdifferentiation and could be an important therapeutic target for the recurrent GBM.

Infiltrating macrophages/microglia in necrotic regions have been linked to poor prognosis via secretion of cytokines (45–48). TNFα, one of the major cytokines secreted by infiltrating macrophages/microglia in the perinecrotic region, induced master TF of the MES transition and promoted conversion from a PN subtype to a MES subtype in an NF-κB signaling-dependent manner (4). However, the molecular link between NF-κB and master TF was unclear. Here, we demonstrated that TNFα-induced NF-κB activation was responsible for the TGM2 expression, and this event ultimately led to MES differentiation via regulation of master TF. Therefore, our results indicate that TGM2 is a molecular mediator between NF-κB and master TF for the necrosis-induced MES transdifferentiation.

Recent studies have introduced an algorithm to identify transcriptional interactions based on GBM expression classes (16). From the transcriptomic findings, several TF, including C/EBPβ, C/EBPδ, TAZ, STAT3, FOSL2, bHLHE40, and RUNX1, were shown to be regulators of the transition to the MES subtype (15, 49). Of these TF, C/EBPβ, TAZ, and STAT3 were identified as master transcriptional regulators. These TF regulate expression of key factors and downstream signaling, including the MES gene signature (15). A high degree of necrosis is observed in the MES class of GBM, and master TF were strongly correlated with the degree of necrosis (3, 15). All three TF are closely involved in self-renewal of MES GSC and maintenance of MES characteristics (49, 50). By blocking TGM2, we effectively downregulated the master TF. Interestingly, TGM2 directly targeted GADD153, a negative regulator of C/EBPβ, to induce protein degradation (40). Our results suggest that TGM2 is a key molecular switch for MES subtype change by regulating master TF.

Radiation-induced necrosis has long been considered a major cause of radioresistance (9, 51). However, there are no clear molecular explanations for this event. In this paper, we demonstrated that irradiation induced TGM2 expression in perinecrotic regions is responsible for the radioresistance by triggering MES subtype change. TGM2 can be a therapeutic biomarker for not only the MES subtype but also non-MES subtypes because therapy-induced necrosis can easily convert non-MES to MES patients. When we compared therapeutic efficacy between MES GSC and non-MES GSC using a combination of radiation and TGM2 silencing, non-MES GSC-injected mice showed better therapeutic benefit following the combination treatment despite their initial low level of TGM2 compared to that of MES GSC. Our results strongly indicate that TGM2 is a therapeutic target that can be used to improve the response to radiotherapy not only MES GBM patients but also for non-MES GBM patients by preventing non-MES to MES transition.

In conclusion, we demonstrated that TGM2 as a key therapeutic target in GBM by regulating the master TF of the MES subtype to increase radioresistance and recurrence. Therefore, our results will provide important therapeutic implication for recurrent GBM.

No potential conflicts of interest were disclosed.

Conception and design: J. Yin, Y.T. Oh, H.-S. Gwak, H. Yoo, S.-H. Lee, J.H. Kim, S.-Y. Kim, D.-H. Nam, J.B. Park

Development of methodology: Y.T. Oh, J.-Y. Kim, S.S. Kim, E. Choi, J.H. Hong, N. Chang, H.J. Kwon, W. Lin, J.H. Kim, S.-Y. Kim, J.B. Park

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N. Chang, S. Park, H.-S. Gwak, D.-H. Nam, M.-J. Park, J.B. Park

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T.H. Kim, N. Chang, H.J. Cho, J.K. Sa, H.-S. Gwak, S.-H. Lee, J.H. Kim, S.-Y. Kim, D.-H. Nam, J.B. Park

Writing, review, and/or revision of the manuscript: J. Yin, Y.T. Oh, J.-Y. Kim, E. Choi, T.H. Kim, H.-S. Gwak, J.H. Kim, M.-J. Park, J.B. Park

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Yin, I. Nakano, J. Lee, J.H. Kim, S.-Y. Kim, D.-H. Nam, J.B. Park

Study supervision: J.H. Kim, S.-Y. Kim, D.-H. Nam, J.B. Park

This research was supported by grants from the National Cancer Center, Republic of Korea (NCC-1410290, NCC-1510061), Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (NRF-2015H1D3A1036090, NRF-2015R1A2A1A15054865, NRF-2015M3A9D9067485, NRF-2017R1A2B4011741, and NRF-2015R1C1A1A01054963), the Korea Institute of Radiological and Medical Science (KIRAMS), funded by the Ministry of Science, ICT and Future Planning, Republic of Korea (1711045557, 1711045538, 1711045554/50531-2017), and the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea, HI14C3418.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Wen
PY
,
Kesari
S
. 
Malignant gliomas in adults
.
N Engl J Med
2008
;
359
:
492
507
.
2.
Stupp
R
,
Hegi
ME
,
Mason
WP
,
van den Bent
MJ
,
Taphoorn
MJ
,
Janzer
RC
, et al
Effects of radiotherapy with concomitant and adjuvant temozolomide versus radiotherapy alone on survival in glioblastoma in a randomised phase III study: 5-year analysis of the EORTC-NCIC trial
.
Lancet Oncol
2009
;
10
:
459
66
.
3.
Verhaak
RG
,
Hoadley
KA
,
Purdom
E
,
Wang
V
,
Qi
Y
,
Wilkerson
MD
, et al
Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1
.
Cancer Cell
2010
;
17
:
98
110
.
4.
Bhat
KP
,
Balasubramaniyan
V
,
Vaillant
B
,
Ezhilarasan
R
,
Hummelink
K
,
Hollingsworth
F
, et al
Mesenchymal differentiation mediated by NF-kappaB promotes radiation resistance in glioblastoma
.
Cancer Cell
2013
;
24
:
331
46
.
5.
Kim
SH
,
Ezhilarasan
R
,
Phillips
E
,
Gallego-Perez
D
,
Sparks
A
,
Taylor
D
, et al
Serine/threonine kinase MLK4 determines mesenchymal identity in glioma stem cells in an NF-kappaB-dependent manner
.
Cancer Cell
2016
;
29
:
201
13
.
6.
Halliday
J
,
Helmy
K
,
Pattwell
SS
,
Pitter
KL
,
LaPlant
Q
,
Ozawa
T
, et al
In vivo radiation response of proneural glioma characterized by protective p53 transcriptional program and proneural-mesenchymal shift
.
Proc Natl Acad Sci U S A
2014
;
111
:
5248
53
.
7.
Mao
P
,
Joshi
K
,
Li
J
,
Kim
SH
,
Li
P
,
Santana-Santos
L
, et al
Mesenchymal glioma stem cells are maintained by activated glycolytic metabolism involving aldehyde dehydrogenase 1A3
.
Proc Natl Acad Sci U S A
2013
;
110
:
8644
9
.
8.
Phillips
HS
,
Kharbanda
S
,
Chen
R
,
Forrest
WF
,
Soriano
RH
,
Wu
TD
, et al
Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis
.
Cancer Cell
2006
;
9
:
157
73
.
9.
Siu
A
,
Wind
JJ
,
Iorgulescu
JB
,
Chan
TA
,
Yamada
Y
,
Sherman
JH
. 
Radiation necrosis following treatment of high grade glioma—a review of the literature and current understanding
.
Acta Neurochir
2012
;
154
:
191
201
.
10.
Raza
SM
,
Lang
FF
,
Aggarwal
BB
,
Fuller
GN
,
Wildrick
DM
,
Sawaya
R
. 
Necrosis and glioblastoma: a friend or a foe? A review and a hypothesis
.
Neurosurgery
2002
;
51
:
2
12
.
11.
Charles
NA
,
Holland
EC
,
Gilbertson
R
,
Glass
R
,
Kettenmann
H
. 
The brain tumor microenvironment
.
Glia
2011
;
59
:
1169
80
.
12.
Schiffer
D
,
Mellai
M
,
Annovazzi
L
,
Caldera
V
,
Piazzi
A
,
Denysenko
T
, et al
Stem cell niches in glioblastoma: a neuropathological view
.
Biomed Res Int
2014
;
2014
:
725921
.
13.
Plaks
V
,
Kong
N
,
Werb
Z
. 
The cancer stem cell niche: how essential is the niche in regulating stemness of tumor cells?
Cell Stem Cell
2015
;
16
:
225
38
.
14.
Hambardzumyan
D
,
Bergers
G
. 
Glioblastoma: defining tumor niches
.
Trends Cancer
2015
;
1
:
252
65
.
15.
Cooper
LA
,
Gutman
DA
,
Chisolm
C
,
Appin
C
,
Kong
J
,
Rong
Y
, et al
The tumor microenvironment strongly impacts master transcriptional regulators and gene expression class of glioblastoma
.
Am J Pathol
2012
;
180
:
2108
19
.
16.
Wang
J
,
Cazzato
E
,
Ladewig
E
,
Frattini
V
,
Rosenbloom
DI
,
Zairis
S
, et al
Clonal evolution of glioblastoma under therapy
.
Nat Genet
2016
;
48
:
768
76
.
17.
Yin
J
,
Park
G
,
Lee
JE
,
Choi
EY
,
Park
JY
,
Kim
TH
, et al
DEAD-box RNA helicase DDX23 modulates glioma malignancy via elevating miR-21 biogenesis
.
Brain
2015
;
138
:
2553
70
.
18.
Yin
J
,
Park
G
,
Lee
JE
,
Park
JY
,
Kim
TH
,
Kim
YJ
, et al
CPEB1 modulates differentiation of glioma stem cells via downregulation of HES1 and SIRT1 expression
.
Oncotarget
2014
;
5
:
6756
69
.
19.
Joo
KM
,
Kim
J
,
Jin
J
,
Kim
M
,
Seol
HJ
,
Muradov
J
, et al
Patient-specific orthotopic glioblastoma xenograft models recapitulate the histopathology and biology of human glioblastomas in situ
.
Cell Rep
2013
;
3
:
260
73
.
20.
Lee
J
,
Kotliarova
S
,
Kotliarov
Y
,
Li
A
,
Su
Q
,
Donin
NM
, et al
Tumor stem cells derived from glioblastomas cultured in bFGF and EGF more closely mirror the phenotype and genotype of primary tumors than do serum-cultured cell lines
.
Cancer Cell
2006
;
9
:
391
403
.
21.
Hoshida
Y
,
Toffanin
S
,
Lachenmayer
A
,
Villanueva
A
,
Minguez
B
,
Llovet
JM
. 
Molecular classification and novel targets in hepatocellular carcinoma: recent advancements
.
Semin Liver Dis
2010
;
30
:
35
51
.
22.
Hoshida
Y
. 
Nearest template prediction: a single-sample-based flexible class prediction with confidence assessment
.
PLoS One
2010
;
5
:
e15543
.
23.
Gautier
L
,
Cope
L
,
Bolstad
BM
,
Irizarry
RA
. 
affy–analysis of Affymetrix GeneChip data at the probe level
.
Bioinformatics
2004
;
20
:
307
15
.
24.
Gentleman
R
,
Carey
V
,
Huber
W
,
Hahne
F
. 
genefilter: genefilter: methods for filtering genes from high-throughput experiments
.
R package version 1.48.1.
; 
2016
.
25.
YiLan
Luo
,
Zeng
RT
. 
clustertend: Check the Clustering Tendency
.
R package version 1.4. Available from
: http://CRAN.R-project.org/package=clustertend; 
2015
.
26.
Kassambara
AM
F. factoextra: extract and visualize the results of multivariate data analyses
.
R package version 1.0.3. Available from:
http://www.sthda.com/english/rpkgs/factoextra; 
2015
.
27.
R Development Core Team
.
R: a language and environment for statistical computing
.
R Foundation for Statistical Computing
.
Available from
: http://www.R-project.org/. 
2015
.
28.
Maechler
M
,
Rousseeuw
P
,
Struyf
A
,
Hubert
M
,
Hornik
K
. 
cluster: cluster analysis basics and extensions
.
R package version 2.0.4. Available from:
https://www.researchgate.net/publication/272176869; 
2016
.
29.
Iismaa
SE
,
Mearns
BM
,
Lorand
L
,
Graham
RM
. 
Transglutaminases and disease: lessons from genetically engineered mouse models and inherited disorders
.
Physiol Rev
2009
;
89
:
991
1023
.
30.
Lorand
L
,
Graham
RM
. 
Transglutaminases: crosslinking enzymes with pleiotropic functions
.
Nat Rev Mol Cell Biol
2003
;
4
:
140
56
.
31.
Ku
BM
,
Kim
DS
,
Kim
KH
,
Yoo
BC
,
Kim
SH
,
Gong
YD
, et al
Transglutaminase 2 inhibition found to induce p53 mediated apoptosis in renal cell carcinoma
.
Faseb J
2013
;
27
:
3487
95
.
32.
Kim
DS
,
Choi
YB
,
Han
BG
,
Park
SY
,
Jeon
Y
,
Kim
DH
, et al
Cancer cells promote survival through depletion of the von Hippel-Lindau tumor suppressor by protein crosslinking
.
Oncogene
2011
;
30
:
4780
90
.
33.
Jang
GY
,
Jeon
JH
,
Cho
SY
,
Shin
DM
,
Kim
CW
,
Jeong
EM
, et al
Transglutaminase 2 suppresses apoptosis by modulating caspase 3 and NF-kappaB activity in hypoxic tumor cells
.
Oncogene
2010
;
29
:
356
67
.
34.
Oh
K
,
Ko
E
,
Kim
HS
,
Park
AK
,
Moon
HG
,
Noh
DY
, et al
Transglutaminase 2 facilitates the distant hematogenous metastasis of breast cancer by modulating interleukin-6 in cancer cells
.
Breast Cancer Res
2011
;
13
:
R96
.
35.
Zoller
M
. 
CD44: can a cancer-initiating cell profit from an abundantly expressed molecule?
Nat Rev Cancer
2011
;
11
:
254
67
.
36.
Brescia
P
,
Richichi
C
,
Pelicci
G
. 
Current strategies for identification of glioma stem cells: adequate or unsatisfactory?
J Oncol
2012
;
2012
:
376894
.
37.
Berezovsky
AD
,
Poisson
LM
,
Cherba
D
,
Webb
CP
,
Transou
AD
,
Lemke
NW
, et al
Sox2 promotes malignancy in glioblastoma by regulating plasticity and astrocytic differentiation
.
Neoplasia
2014
;
16
:
193
206
.
38.
Yin
J
,
Park
G
,
Kim
TH
,
Hong
JH
,
Kim
YJ
,
Jin
X
, et al
Pigment epithelium-derived factor (PEDF) expression induced by EGFRvIII promotes self-renewal and tumor progression of glioma stem cells
.
PLoS Biol
2015
;
13
:
e1002152
.
39.
Ron
D
,
Habener
JF
. 
CHOP, a novel developmentally regulated nuclear protein that dimerizes with transcription factors C/EBP and LAP and functions as a dominant-negative inhibitor of gene transcription
.
Genes Dev
1992
;
6
:
439
53
.
40.
Batchvarova
N
,
Wang
XZ
,
Ron
D
. 
Inhibition of adipogenesis by the stress-induced protein CHOP (Gadd153)
.
EMBO J
1995
;
14
:
4654
61
.
41.
Bao
S
,
Wu
Q
,
McLendon
RE
,
Hao
Y
,
Shi
Q
,
Hjelmeland
AB
, et al
Glioma stem cells promote radioresistance by preferential activation of the DNA damage response
.
Nature
2006
;
444
:
756
60
.
42.
Baumann
M
,
Krause
M
,
Hill
R
. 
Exploring the role of cancer stem cells in radioresistance
.
Nat Rev Cancer
2008
;
8
:
545
54
.
43.
Ku
BM
,
Kim
SJ
,
Kim
N
,
Hong
D
,
Choi
YB
,
Lee
SH
, et al
Transglutaminase 2 inhibitor abrogates renal cell carcinoma in xenograft models
.
J Cancer Res Clin Oncol
2014
;
140
:
757
67
.
44.
Kang
JH
,
Lee
JS
,
Hong
D
,
Lee
SH
,
Kim
N
,
Lee
WK
, et al
Renal cell carcinoma escapes death by p53 depletion through transglutaminase 2-chaperoned autophagy
.
Cell Death Dis
2016
;
7
:
e2163
.
45.
Lewis
C
,
Murdoch
C
. 
Macrophage responses to hypoxia: implications for tumor progression and anti-cancer therapies
.
Am J Pathol
2005
;
167
:
627
35
.
46.
Murdoch
C
,
Giannoudis
A
,
Lewis
CE
. 
Mechanisms regulating the recruitment of macrophages into hypoxic areas of tumors and other ischemic tissues
.
Blood
2004
;
104
:
2224
34
.
47.
Tripathi
C
,
Tewari
BN
,
Kanchan
RK
,
Baghel
KS
,
Nautiyal
N
,
Shrivastava
R
, et al
Macrophages are recruited to hypoxic tumor areas and acquire a pro-angiogenic M2-polarized phenotype via hypoxic cancer cell derived cytokines Oncostatin M and Eotaxin
.
Oncotarget
2014
;
5
:
5350
68
.
48.
Leek
RD
,
Landers
RJ
,
Harris
AL
,
Lewis
CE
. 
Necrosis correlates with high vascular density and focal macrophage infiltration in invasive carcinoma of the breast
.
Br J Cancer
1999
;
79
:
991
5
.
49.
Bhat
KP
,
Salazar
KL
,
Balasubramaniyan
V
,
Wani
K
,
Heathcock
L
,
Hollingsworth
F
, et al
The transcriptional coactivator TAZ regulates mesenchymal differentiation in malignant glioma
.
Genes Dev
2011
;
25
:
2594
609
.
50.
Carro
MS
,
Lim
WK
,
Alvarez
MJ
,
Bollo
RJ
,
Zhao
X
,
Snyder
EY
, et al
The transcriptional network for mesenchymal transformation of brain tumours
.
Nature
2010
;
463
:
318
25
.
51.
Debus
J
,
Abdollahi
A
. 
For the next trick: new discoveries in radiobiology applied to glioblastoma
.
Am Soc Clin Oncol Educ Book
2014
:
e95
9
.