Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 May 9.
Published in final edited form as: Cancer Cell. 2016 May 9;29(5):669–683. doi: 10.1016/j.ccell.2016.03.027

Olig2-Dependent Reciprocal Shift in PDGF and EGF Receptor Signaling Regulates Tumor Phenotype and Mitotic Growth in Malignant Glioma

Fanghui Lu 1,2, Ying Chen 2,3, Chuntao Zhao 2, Haibo Wang 2, Danyang He 4, Lingli Xu 2, Jincheng Wang 2, Xuelian He 2, Yaqi Deng 2, Ellen E Lu 2, Xue Liu 3, Ravinder Verma 2, Hong Bu 1, Rachid Drissi 2, Maryam Fouladi 2, Anat O Stemmer-Rachamimov 5, Dennis Burns 4, Mei Xin 2, Joshua B Rubin 6, El Mustapha Bahassi 7, Peter Canoll 8, Eric C Holland 9, Q Richard Lu 2,10,#
PMCID: PMC4946168  NIHMSID: NIHMS774896  PMID: 27165742

Summary

Malignant gliomas exhibit extensive heterogeneity and poor prognosis. Here we identify mitotic Olig2-expressing cells as tumor-propagating cells in proneural gliomas, elimination of which blocks tumor initiation and progression. Intriguingly, deletion of Olig2 resulted in tumors that grow, albeit at a decelerated rate. Genome occupancy and expression profiling analyses reveal that Olig2 directly activates cell proliferation machinery to promote tumorigenesis. Olig2 deletion causes a tumor phenotypic shift from an oligodendrocyte precursor-correlated proneural toward astroglia-associated gene expression pattern, manifest in down-regulation of PDGF receptor-alpha and reciprocal up-regulation of EGFR. Olig2 deletion further sensitizes glioma cells to EGFR inhibitors and extends animal lifespans. Thus, Olig2-orchestrated receptor signaling drives mitotic growth and regulates glioma phenotypic plasticity. Targeting Olig2 may circumvent resistance to EGFR-targeted drugs.

Graphical Abstract

graphic file with name nihms774896f9.jpg

eTOC

Lu et al. show that ablation of dividing Olig2-expressing cells in a glioma model reduces tumor initiation and growth. Olig2 deletion in glioma also delays growth, changes the gene expression profile from proneural to classical subtype, and leads to increased EGFR expression and sensitivity to EGFR inhibitors.

Introduction

Glioblastoma (GBM) is the most common malignant brain tumor in adults, exhibiting distinct molecular characteristics, and patients have very poor prognosis with a median survival of less than one year despite aggressive treatments (Jansen et al., 2010). Tumors are resistant to conventional radiotherapy and chemotherapies, and the efficacy of current treatments is limited (Ohgaki and Kleihues, 2005; Schonberg et al., 2014). Based on gene expression profiles, GBMs have been classified into four distinct molecular subtypes, namely proneural, classical, neural, and mesenchymal with distinct gene expression signatures (Verhaak et al., 2010). The proneural subtype is highly enriched with the signature associated with oligodendrocyte lineage cells, whereas the classical subgroup is strongly associated with the astrocytic signature, and the mesenchymal subgroup is enriched with a gene signature associated with cultured/reactive astrocytes and microglia (Lei et al., 2011; Verhaak et al., 2010). Much of the heterogeneity of GBMs can be attributable to their distinct genetic alterations (Brennan et al., 2013; Carro et al., 2010). The proneural subtype displays characteristic genetic alterations including PDGFRA amplification and TP53 mutations, as well as IDH1 or IDH2 mutations (Brennan et al., 2013; Verhaak et al., 2010), while the classical subtype is characterized by mutational activation EGFR or by extra copies of EGFR (Hayden, 2010). Although distinct events occurring in different target cells likely contribute to the variety of GBM phenotypes, the molecular determinants that regulate the tumor phenotype are not fully understood.

Depending on genetic alterations, glioma cells may transition between different states by utilizing alternative pathways that incite tumor growth and progression (Johnson et al., 2014; Meacham and Morrison, 2013). Since either activation of TNF-α/NF-κB or loss of NF1 converts proneural GBM to the mesenchymal subtype (Bhat et al., 2013; Ozawa et al., 2014), GBM tumor cells therefore manifest phenotypic plasticity. This plasticity may render tumor cells more invasive or resistant to current therapies at different stages in their development (Friedmann-Morvinski et al., 2012; Persson et al., 2010). At present, the underlying genetic alterations and the signaling mechanisms that result in transitions between different tumor cell states remain elusive. Identification of the molecular control of tumorigenic cell properties and cellular hierarchies within GBM are essential for understanding pathogenic processes and may lead to potential avenues for targeted GBM treatment, especially with regard to confronting resistance.

Recent studies indicate that a population of stem-like tumor propagating cells appears to drive tumor growth and progression in GBM (Chen et al., 2012; Liu et al., 2011; Schonberg et al., 2014). OLIG2, an early marker for oligodendroglial lineage progenitors (Lu et al., 2002), is expressed in all grades of diffuse gliomas (Ligon et al., 2004). Remarkably, the proneural tumor subtype possesses a gene expression profile that resembles that of oligodendrocyte precursor cells (OPCs) (Lei et al., 2011; Liu et al., 2011; Verhaak et al., 2010), a presumptive cell type of origin for this type of GBM. Moreover, OLIG2 has been identified as one of core transcription factors that reprogram differentiated GBM cells into the stem-like propagating cells (Suva et al., 2014). Previous studies indicate that neural progenitors isolated from Olig1/2−/−;Cdkn2a−/− embryos cannot be induced by oncogenic EGFRvIII to form glioma in murine allografts (Ligon et al., 2007). In addition, Olig2 appears to directly oppose p53 responses to genotoxic damage to regulate glioma growth (Mehta et al., 2011). Currently, it is not clear whether Olig2 is required for endogenous glioma formation, particularly on a Trp53-mutant background, since p53 is one of the most frequently inactivated proteins in the development of malignant glioma or GBM in humans (Verhaak et al., 2010). The molecular mechanisms whereby Olig2 regulates glioma tumorigenesis are not fully understood. Furthermore, because Olig2+ cells are present in most diffuse gliomas, whether Olig2+ propagating cells are critical for directing tumor cell growth and progression in GBM remains unknown.

Results

Intense Olig2 expression characterizes mitotic progenitors in human and mouse gliomas

To characterize OLIG2 expression in mitotic progenitors in human proneural-like GBM with PDGFRA amplification, we performed immunostaining for OLIG2 and a proliferative marker, Ki67. We detected extensive OLIG2 expression in tumor lesions (Figure 1A). Approximately 35 ± 5 % of OLIG2+ cells expressed Ki67 among the GBMs examined (Figures 1B and 1C), and substantial populations of OLIG2+ cells were co-labeled with SOX2, POU3F2, or CD133 (Figures 1B and 1C), the markers for tumor initiating/propagating cells (Schonberg et al., 2014). These tumor propagation-associated markers were enriched on OLIG2+ cell populations in GBM lesions (Figures S1A and S1B). Similarly, a large population of Ki67+ cells expressed OLIG2 in proneural GBM (Figure S1C), which is consistent with previous findings (Ligon et al., 2007). These observations suggest OLIG2+ cells are highly proliferative with tumor progenitor properties in proneural-like GBM lesions.

Figure 1. Olig2 expression in mitotic progenitors in human and mouse GBM.

Figure 1

Open in a new tab

(A) A representative image showing OLIG2 (brown) expression in human proneural (PN) GBMs.

(B) Human PN GBM sections immunostained with OLIG2 (red) and Ki67, SOX2, POU3F2, or CD133 as indicated (green). Arrows indicate co-labeled cells.

(C) The percentage of labeling-positive cells among OLIG2+ cells in proneural GBMs (n = 6 individual tumors; > 250 cell counts/tumor tissue).

(D) Left: PDGFB-Cre retrovirus injection into the cerebral white matter of 8-week old adult Ptenfl/fl;Trp53fl/fl mice. Right: Hematoxylin and eosin (H&E)-stained brain section showing malignant glioma at dpi 24.

(E) Olig2 expression (brown) in PDGFB-Cre-induced glioma lesion.

(F–H) Glioma lesions were immunostained with Olig2 (red) and (F) Ki67, (G) Sox2, or (H) Pou3f2 (green).

(I) The percentage of co-labeled cells among Olig2+ cells in tumor lesions. (n = 5 animals; > 250 cell counts/tumor tissue).

Scale bars in A–B, E–H, 50 µm; D, 1 mm. Data are presented as mean ± S.E.M. in C and I. See also Figure S1.

To investigate the role of Olig2+ cells in glioma formation, we induced malignant gliomas in adult mice by deleting both floxed alleles of Pten and Trp53 with a Cre-expressing retrovirus carrying PDGFB through stereotaxic microinjection into the cerebral white matter (Figure 1D) (Lei et al., 2011). This murine model of malignant glioma closely resembles the human proneural GBM (Lei et al., 2011). These tumors carry well-defined genetic alterations with PDGF pathway activation accompanied by Trp53 and Pten losses, exhibiting pathological and morphological characteristics of human GBM (Figure S1D) (Lei et al., 2011; Sonabend et al., 2014). These mouse gliomas consisted of a large population of Olig2-expressing cells (Figure 1E), which were highly proliferative as indicated by Ki67 expression in approximately 46 ± 6 % of Olig2+ cells (Figures 1F and 1I). The proportion of Ki67+ proliferative cells among Olig2+ cells was higher than that in Olig2− cells (Figures 1I and S1E). Similar to human proneural GBM, a substantial population of Olig2+ cells expressed Sox2 and Pou3f2 (Figures 1G–1I and S1E). In addition, a large majority of Ki67+ cells were Olig2-positive (Figure S1F), suggesting that Olig2+ mitotic cells are a major component of the proliferating cell mass in this mouse glioma model.

Ablation of mitotic Olig2+ cells inhibits glioma formation and progression

To test whether Olig2+ proliferative cells are the critical source of glioma growth, we utilized an in vivo cell suicide approach to deplete mitotic Olig2+ progenitors by generating Olig2-TK (thymidine kinase) mice with the ganciclovir (GCV)-inducible suicide gene HSV-TK knocked-in at the Olig2 locus (Figures 2A and 2B). TK expression in Olig2-TK knock-in mice mirrored Olig2 expression patterns (Figure 2C). HSV-TK kinase activity converts GCV into toxic triphosphates that inhibit DNA polymerase and eliminate actively dividing tumor cells, while sparing normal post-mitotic Olig2+ cells like mature oligodendrocytes (Figures 2D, S2A and S2B). When GCV was administered to the glioma-forming mice carrying Olig2-TK at day 5 post PDGFB-Cre virus injection (dpi 5), tumor growth was markedly inhibited compared with control tumor-forming animals (Ptenfl/fl;Trp53fl/fl:PDGFB-Cre; designated as Ctrl-T) (Figures 2E and 2F). GCV-treated Olig2-TK mice had a significantly extended survival curve relative to that of control tumor mice (Figure 2E). Histologic analysis revealed that the majority of the Olig2-TK mice had no detectable tumor mass and exhibited few Ki67+ proliferative cells in the brain compared with Ctrl-T mice (Figures 2F and 2G), suggesting that Olig2+ mitotic cells represent a population of tumor cells that are necessary for tumor propagation in this animal model.

Figure 2. Ablation of mitotic Olig2+ cells inhibits glioma formation.

Figure 2

Open in a new tab

(A) Schematic design for Olig2-TK knock-in mice. The line below the Olig2 locus represents the 5’ external probe for Southern analysis. H, HindIII site; neo, neomycin cassette; Frt, flippase recognition targets; DTA, diphtheria toxin gene.

(B) Southern blot analysis of Olig2-TK knock-in with 5’ Olig2 probe using HindIII digested DNA. The 3.2 and 5-kb bands correspond to the knock-in and wild-type alleles, respectively.

(C) Immunostaining of TK and Olig2 in the cortex of P14 Olig2-TK mice. Arrows: co-labeled cells.

(D) Schematic diagram showing GCV-mediated depletion of dividing Olig2-TK+ cells that spares post-mitotic Olig2+ cells.

(E) Kaplan-Meier survival analysis of GCV-treated control (n = 13) and Olig2-TK (n = 14) mice from dpi 5 (*** p < 0.001 with the log-rank test).

(F) H&E-stained brain sections from GCV-treated control and Olig2-TK mice. Arrows: tumor lesions.

(G) H&E and Ki67 staining of GCV-treated Ctrl-T mice (top panels) and Olig2-TK mice (bottom panels).

(H) Survival analysis of mice treated with GCV starting from dpi 20. Median survival in control group 25 days (n = 7) and in Olig2-TK group (n = 8) 39.5 days (** p < 0.01 with the log-rank test).

(I, J) The percentage of Olig2+/Ki67+ cells among Olig2+ cells (I) and of Olig2−/Ki67+ cells among Ki67+ cells (J) in control tumors at dpi 5 and dpi 23. Data represent the means ± SEM in tumor tissues from three animals (> 250 cell counts/tumor tissue; * p < 0.05; Student’s t test).

Scale bars in C and G, 50 µm; F, 1 mm. See also Figure S2.

Even when GCV treatment was performed at dpi 20, a late phase of tumorigenesis, the gliomaforming Olig2-TK mice survived longer compared to control groups (Figure 2H). However, the survival rate was much reduced in the animal group administered GCV at dpi 20 compared to those treated earlier in tumorigenesis at dpi 5 (Figures 2E and 2H). We observed that the proportion of proliferative Olig2+ cells decreased while the percentage of Olig2− cells among Ki67+ cells increased in Ctrl-T tumors at dpi 23 compared with dpi 5 (Figures 2I and 2J). The increase in Olig2− proliferative cells suggests that both Olig2+ and Olig2− mitotic cells are responsible for the tumor growth during the late stage of tumor progression.

Olig2 deletion delays the growth and progression of glioma

To determine the function of Olig2 in glioma formation, we bred floxed Pten and Trp53 mice with mice carrying an Olig2 floxed allele (Yue et al., 2006). Cre virus transduction will ablate the Olig2fl/fl allele together with Pten and Trp53 floxed alleles in the same cells. The resulting animals with Olig2 deletion (Ptenfl/fl;Trp53fl/fl;Olig2fl/fl:PDGFB-Cre) are designated as Olig2cKO. In contrast to robust Olig2 expression in Ctrl-T tumors, Olig2 was essentially undetectable in Cre-expressing cells in Olig2cKO tumors (Figure S3A). Residual Olig2+ cells observed in the Olig2cKO tumors were not Cre-retrovirus transduced (Figure S3A), and therefore they do not have Cre-mediated Pten and Trp53 deletion; these cells might represent non-transformed OPCs that have been entrapped or recruited to the tumor.

To trace the growth and progression of tumors from Cre-recombined cells, we performed in vivo bioluminescence imaging of the Ctrl-T and Olig2cKO mice carrying the Rosa26-tmLuciferase reporter (Figure 3A). In the Ctrl-T mice with Pten and Trp53 deletion, initial tumor formation was detected around dpi 15. Tumors grew rapidly and spread extensively in the brain over time as indicated by increasing luminescent signal (Figure 3A). Most control tumor-forming mice died before dpi 30 (Figure 3A). In contrast, in the Olig2cKO mice, we detected tumor appearance at a later stage around dpi 20. Intriguingly, tumor mass continued to expand before most Olig2cKO mice succumbed to the tumor by around dpi 50 (Figures 3A and 3B), indicating that Olig2 ablation led to a slower rate of growth and/or a delay in tumor initiation. However, the size of tumors monitored by luminescent signals gradually increased and eventually reached a level resembling the control tumor groups at the terminal stage of tumor growth (Figure 3B).

Figure 3. Olig2 deletion inhibits the growth of mouse glioma.

Figure 3

Open in a new tab

(A) Representative bioluminescence images of Ctrl-T and Olig2cKO mice during tumor progression.

(B) Quantification of bioluminescence signals as a function of time for Ctrl-T and Olig2cKO mice. Data are the means ± SEM (n = 6 animals).

(C) H&E staining of brain sections of Ctrl-T and Olig2cKO tumors at dpi 13, 20, and 48.

(D) Sections of Ctrl-T and Olig2cKO tumors at dpi 13 were immunostained with Olig2, BrdU, and Sox2 after a 2-hr BrdU pulse labeling.

(E) Sections of Ctrl-T and Olig2cKO tumors at dpi 10 and dpi 20 immunostained for BrdU (green) after a 2-hr BrdU pulse labeling. DAPI in blue.

(F) Percentage of BrdU+ cells in Ctrl-T and Olig2cKO tumors at dpi 10 and dpi 20. Data represent the means ± SEM from three independent experiments (** p < 0.01; Student’s t test).

(G) Survival analysis of Ctrl-T and Olig2cKO mice. Median survival in Ctrl-T group (n = 18): 25 days; in Olig2cKO group (n = 37): 48 days (*** p < 0.001 with the log-rank test).

Scale bars in C, 1 mm; D, 20 µm. E, 50 µm. See also Figure S3.

Histological analyses of tumors confirmed glioma formation and growth across time-points post-viral injection in Ctrl-T and Olig2cKO mice (Figures 3C, S3B and S3C). The glioma in Olig2cKO mice in the end-stage tumors at dpi 40 was similar to that of control tumors (Figures S3C–S3E), resembling human GBM with characteristic high mitotic index and pseudopalisading necrosis. To examine the effect of Olig2 ablation on tumor cell proliferation, we carried out BrdU pulse labeling of proliferating cells. In the Olig2cKO tumors, Olig2 expression in the neoplasm was essentially absent, although Sox2+/BrdU+ proliferative cells remained in the lesion (Figure 3D). The residual Olig2+ cells expressed neither Sox2 nor BrdU, suggesting that they are not the major contributors to the continued, albeit slower, tumor growth in Olig2cKO mice (Figure 3D). The number of BrdU+ cells was significantly reduced in the Olig2cKO tumors during the early stage of tumorigenesis (Figures 3E and 3F). In late-stage tumors, however, the BrdU labeling was comparable between the two groups (Figures 3E and 3F). Kaplan-Meier survival analysis indicated that Olig2 deletion resulted in prolonged survival (Figure 3G), suggesting that Olig2 deletion delays, but does not prevent, glioma initiation and progression.

Olig2-deleted tumor cells exhibit markedly reduced sphere-forming capacity but retain tumorigenic capacity

To examine the self-renewal capacity of tumor cells in the absence of Olig2, we performed neurosphere formation assays. In contrast to Ctrl-T counterparts, which formed free-floating spheres, the Olig2-deleted tumor cells readily attached to form a monolayer culture (Figure 4A). Primary spheres were then dissociated and plated at clonal density over serial passages. The control tumor cells maintained their sphere formation capacity, whereas very few spheres were formed by Olig2cKO tumor cells during passages (Figures 4A and 4B). When tumor cells isolated from Ctrl-T and Olig2cKO neoplasms were plated as adherent monolayers in the same mitogenic conditions, the Olig2-deleted tumor cells were able to expand and proliferate, as assessed by BrdU incorporation, although these cells exhibited a lower proliferation rate than control tumor cells (Figures 4C and 4D).

Figure 4. Proliferation and differentiation of Olig2cKO tumor cells.

Figure 4

Open in a new tab

(A) Neurosphere formation from Ctrl-T and Olig2cKO tumor cells in serum-free media. Scale bar, 50 µm.

(B) Percentage of wells that formed spheres from Ctrl-T and Olig2cKO tumor cells at clonal density in continuous passages.

(C) Ctrl-T and Olig2cKO tumor cell cultures were pulse-labeled with BrdU for 1 hr and immunostained with BrdU and Olig2 and counterstained with DAPI. Scale bar, 20 µm.

(D) Percentage of BrdU+ Ctrl-T and Olig2cKO tumor cells pulse-labeled with BrdU for 1 hr.

(E) Representative images of soft agar colony formation by Ctrl-T and Olig2cKO tumor cells. Scale bar, 1 mm.

(F, G) Quantification of colony diameter (F) and colony number (G) from Ctrl-T and Olig2cKO tumor cells, respectively, plated on soft agar.

(H) Tumor cells from Ctrl-T and Olig2cKO mice were cultured in 1% FBS for 48 hr and stained with anti-GFAP and anti-PDGFRα as indicated. Scale bar, 50 µm.

Data are presented as means ± SEM from three independent experiments in B, D, F and G. The cross-lines in F represent means ± SEM (** p < 0.01; *** p < 0.001; Student’s t test). See also Figure S4.

We further analyzed the clonogenic capacity of Olig2 mutant cells using the soft agar assay at clonal density. Ctrl-T and Olig2cKO tumor cells were able to form colonies and displayed similar clonogenicity; however, the overall sizes of colonies formed by Olig2cKO tumor cells were smaller than those of Ctrl-T cells over 20 days (Figures 4E–4G), suggesting a slower growth rate in Olig2cKO than control tumor cells. When cultured under neural differentiation conditions, Olig2+ tumor spheres differentiated into PDGFRα+ OPC-like cells along with a few GFAP-expressing cells (Figure 4H); in contrast, the vast majority of Olig2-deleted cells were predominantly differentiated into GFAP+ astrocyte-like cells but not PDGFRα+ OPC-like cells (Figure 4H), indicating that Olig2 is required for PDGFRα expression and maintenance of growth of PDGFRα+ OPC-like cells, while suppressing astrocytic GFAP expression. Similarly, we detected an up-regulation of GFAP in Olig2cKO tumors (Figure S4A). Olig2 deletion resulted in an increase of astroglial gene expression while causing a down-regulation of OPC-associated gene expression in cultured tumor cells (Figure S4B). This is consistent with Olig2 function in control of oligodendrocyte and astrocyte fate switch (Cai et al., 2007; Zhu et al., 2012).

To investigate the tumorigenic capacity of the cells derived from Olig2-deleted tumors in vivo, we performed orthotopic transplantation into the cerebral striatum of immunodeficient NOD scid gamma (NSG) mice, which were engrafted with 1 × 104 or 1 × 103 tumor cells. The cells transplanted from either Ctrl-T or Olig2cKO tumors formed secondary tumors that resembled their parental tumors histologically (Figure 5A). Kaplan-Meier survival analysis revealed prolonged survival of mice with secondary tumors derived from Olig2cKO tumors compared with those derived from Ctrl-T tumors (Figure 5B). To determine the growth rate of Olig2cKO tumor cells in the engrafted mice, we monitored bioluminescence for tumor development and found that Olig2-deletion stalled progression of tumorigenesis (Figure 5C), paralleling results in Olig2cKO primary tumors. Similar to primary tumors, the residual Olig2+ cells in transplanted Olig2cKO tumors were not proliferative (Figures 5D and 5E) and were without Cre expression (data not shown), suggesting that these cells were non-transformed Olig2+ cells residing within tumor lesions. Together, our data indicate that Olig2-deleted tumor cells maintain tumorigenic capacity despite a relatively slower growth rate.

Figure 5. Olig2cKO tumor cells retain tumorigenic capacity.

Figure 5

Open in a new tab

(A) Tumor cells (1 × 104) isolated from Ctrl-T and Olig2cKO tumor lesions were intracranially engrafted into the striatum of NSG mice. The tumor tissues were stained with H&E and Olig2 as indicated. Arrows indicate tumor tissues. Scale bars in left, middle and right panels; 1 mm, 100 µm and 50 µm, respectively.

(B) Kaplan-Meier survival analysis of NSG mice transplanted with 1 × 104 Ctrl-T or Olig2cKO tumor cells or 1 × 103 Ctrl-T or Olig2cKO tumor cells (p < 0.01 with the log-rank test between control and Olig2cKO group; n ≥ 5 for each group).

(C) Representative in vivo bioluminescence imaging of the NSG mice engrafted with 1 × 104 Ctrl-T or Olig2cKO tumor cells carrying the Rosa-tmLuciferase reporter at dpi 16 and dpi 30.

(D) Tumor lesions after transplantation of Ctrl-T or Olig2cKO tumor cells were immunostained with Olig2 (red) and BrdU (green) after 2-hr pulse-labeling. Arrows indicate Olig2+ cells. Scale bar, 20 µm.

(E) Percentage of BrdU+ cells among Olig2+ or Olig2− cells in Ctrl-T and Olig2cKO orthotopic tumors. Data represent the means ± SEM from three mice per group.

Olig2 directly targets the enhancers of cell cycle regulators and oncogenes to regulate their expression

To determine the mechanisms for Olig2 regulation of tumor growth, we carried out transcriptome profiling of control and Olig2cKO tumor tissues as well as normal adult cortical tissues using RNA-sequencing. By interrogating gene expression signatures from the Molecular Signatures Database (MSigDB) (Subramanian et al., 2005), we identified the pathways associated with tumorigenesis including DNA replication, oncogenic pathways and cell proliferation were substantially activated in Ctrl-T tumors comparing with normal cortex, while those pathways activities were downregulated in Olig2cKO tumors (Figure 6A). Similarly, the genes down-regulated in Olig2cKO tumors as compared to Ctrl-T tumors were related to the regulators of cell proliferation, survival, stemness and oncogenic programs (Figures 6B and 6C). These observations suggest that Olig2 promotes tumorigenesis through controlling multiple oncogenic pathways.

Figure 6. Olig2 targets the enhancers of cell cycle regulators and oncogenes to regulate cell growth.

Figure 6

Open in a new tab

(A) Heatmap showing the patterns of pathway activity among normal cortices (n=3), Ctrl-T tumors (n = 4) and Olig2cKO tumors (n = 6) based on the MSigDB database. Expression signature scores are the means and clustered with linkage hierarchical clustering.

(B) Volcano plot of transcriptome expression profiles between the Ctrl-T tumors and Olig2cKO tumors. Red and green dots represent genes significantly up-regulated and downregulated in Olig2cKO tumors (p < 0.05), respectively.

(C) Heatmap of the genes differentially expressed and associated with tumorigenesis between Ctrl-T tumors (n = 4) and Olig2cKO tumors (n = 6).

(D) Anchor plot of Olig2 and H3K27ac ChIP-seq peaks.

(E) Heatmap for signal intensity of Olig2-targeted gene loci in Ctrl-T tumors and NPCs.

(F) Venn diagram for Olig2-targeted genes detected by ChIP-seq in tumor tissues and NPCs.

(G) GO analysis of biological processes corresponding to Olig2-targeted genes in tumor tissues. (H) Olig2 and histone binding profiles on indicated gene loci in tumor tissues.

Olig2 targeting in NPCs and OPCs is shown in the bottom panels.

(I) qRT-PCR analysis of candidate gene expression in the wild-type corpus callosum (WT-CC) at 8-week old and Ctrl-T tumor tissues at dpi 20.

(J) qRT-PCR analysis of gene expression in Ctrl-T and Olig2cKO tumor tissue.

(K) qRT-PCR quantification of c-Myc and c-Jun expression in an early stage (dpi 10) Ctrl-T and Olig2cKO tumors.

Data are presented as the means ± SEM from three animals per group in I, J, K (* p < 0.05; ** p < 0.01; *** p < 0.001; Student’s t test).

To identify the genes directly regulated by Olig2, we performed whole-genome chromatin immunoprecipitation sequencing (ChIP-seq) using Olig2-expressing control tumors. The Olig2-targeted genes were highly enriched in the enhancer regions, marked by the activating histone mark H3K27ac, in the genome of tumor tissues (Figure 6D). When comparing the Olig2 occupancy profiles with those of Olig2-enriched neural progenitor cells (NPCs) (Meijer et al., 2014), we detected a unique subset of gene promoters/enhancers exhibiting strong Olig2 enrichment in the neoplasm but not in normal NPCs (Figure 6E). A cohort of Olig2-targeted sites was identified specifically in tumors (Figure 6F), but not in Olig2-enriched NPCs or OPCs (Yu et al., 2013). These sites are mainly associated with genes known to encode regulators of cell growth and proliferation such as cell cycle regulators Cdca8, Cdc20, Cdc25c, along with cell proliferation-promoting factors Aurka and Pim1 (Figures 6G and 6H). Strikingly, we detected a strong enrichment of Olig2 binding in the enhancer/promoter regions of oncogenic genes, marked by H3K27ac and H3K4me3, such as proto-oncogenes Myc and Jun (Figure 6H). Expression of these genes is required for maintenance of glioma stem cell growth (Wang et al., 2008), and has been shown to promote cell cycle progression and cellular proliferation (Nie et al., 2012).

To further confirm that expression of these target genes was dependent on Olig2 expression in the tumors, we performed qPCR analysis and found that expression of the Olig2-targeted genes such as Cdc20, Cdc25c, Aurka and Myc were dramatically increased in the Olig2-expressing Ctrl-T tumors compared with the wild-type corpus callosum (Figure 6I). By contrast, transcripts from these cell cycle-promoting genes and proto-oncogenes c-Myc and c-Jun were down-regulated in the Olig2cKO tumors compared to Ctrl-T tumors (Figures 6J and 6K), suggesting that Olig2 transcriptionally targets the enhancers of the genes encoding the regulators of cell growth, and activates their expression to promote cell proliferation during tumorigenesis.

Loss of Olig2 leads to a proneural phenotype shift to an astroglia-associated classical-like gene expression pattern

Given the persistent, but delayed, glioma formation in Olig2cKO mice, we questioned whether alternative pathways sustain the growth and progression of Olig2-deleted tumors. We therefore compared the transcriptome profiles of Ctrl-T and Olig2cKO tumors. We compared these profiles to those of validated subtypes of human GBM samples from the Cancer Genome Atlas Project (CGAT, 2008). Gene regulatory network analysis revealed that the genes that were highly enriched in Ctrl-T tumors were closely related to oligodendrogenesis and cell proliferation, and with those associated with the proneural subtype of GBM characterized by an oligodendrocytic gene expression pattern (Figures 7A and 7B). In contrast, the genes that were over-represented in Olig2cKO tumors were correlated with astrogliogenesis and resembled the gene expression pattern of classical-like tumors (Figure 7B). These observations suggest that Olig2 deletion leads to a tumor phenotype shift. By analyzing the expression level of signature genes identified in different GBM subtypes (Verhaak et al., 2010) between Ctrl-T and Olig2cKO tumors, we found that Olig2-deletion led to downregulation of proneural signature genes and up-regulation of a set of classical signature genes (Figure 7C). Consistently, the gene sets enriched in Ctrl-T and Olig2cKO tumors were over-represented in proneural and classical GBMs, respectively (Figure 7D).

Figure 7. Olig2 deletion leads to a tumor phenotype shift from proneural to classical expression pattern.

Figure 7

Open in a new tab

(A) Scatter plot of RNA-seq data from Ctrl-T and Olig2cKO tumors. Examples of significantly (p < 0.05) up-regulated and down-regulated genes in Olig2cKO as compared to Ctrl-T are represented by red and blue dots, respectively.

(B) ToppCluster analysis shows associated biological processes of genes enriched in Ctrl-T and Olig2cKO tumors.

(C) Heatmap indicating expression levels of the “signature” genes in PN and CL GBMs (Verhaak et al., 2010) that are substantially altered in Olig2cKO tumors as compared to Ctrl-T tumors.

(D) Heatmap of the Ctrl-T-enriched genes and Olig2cKO-enriched genes in four subtypes of human GBM taken from TCGA unified CORE datasets.

(E) GSEA enrichment plots showing the comparison of gene expression profiles in Ctrl-T and Olig2cKO tumors with the TCGA signature gene sets in PN, CL, mesenchymal (MES) or neural (NL) GBMs as indicated. NES: normalized enrichment score; p value, represents the statistical significance of the enrichment score; FDR: false discovery rate.

(F) qRT-PCR analysis of expression of representative PN, CL, MES, and NL signature genes between Ctrl-T and Olig2cKO tumors.

(G) Visualization of Olig2 binding profiles on the enhancers (marked by H3K27ac) of representative CL signature genes, Egfr, Tbx2, and Gli2 in Ctrl-T and Olig2cKO tumors.

(H) Tumor cells from three different Ctrl-T mice were infected with lentivirus for expression of an shRNA targeting Olig2 or with an shRNA non-target (NT) control for 4 days. The expression of a set of PN and CL genes was analyzed by qRT-PCR.

(I) Graph of the relative OLIG2 or EGFR expression (Z-score) in PN and CL subtypes of GBMs from the core TCGA samples using the unified scaled data (*** p < 0.001; Student’s t test).

(J) Graph of the relative OLIG2 and EGFR expression in PN or CL subtypes of core TCGA GBMs, respectively.

(K) Primary human GBM clonal cells TS543 were transduced with lentiviral non-target or OLIG2 shRNA vectors carrying a GFP reporter for 72 hr and were pulse-labeled with BrdU for 1 hr. Representative images of the cells immunostained with anti-BrdU and OLIG2 are shown. Arrows: transduced cells. Scale bar, 10 µm.

(L) Percentage of BrdU+ cells transduced with non-target (NT) control shRNA and OLIG2 shRNA in TS543 (left) and TS667 (right).

(M) Primary human proneural GBM clonal cells TS543 (left) and TS667 (right) were infected with lentivirus for non-target (NT) control or OLIG2 shRNA for 84 hr. Expression of a set of proneural and classical tumor associated genes were analyzed by qRT-PCR.

Data are presented as the means ± SEM from three independent experiments in F, H, L, and M. The cross-lines in I and J represent means ± SEM (* p < 0.05; ** p < 0.01; *** p < 0.001; Student’s t test). See also Figure S5.

To define potential glioma phenotype shift in an unbiased manner, we then performed Gene Set Enrichment Analysis (GSEA) (Subramanian et al., 2005) to compare the expression pattern of Ctrl-T and Olig2cKO tumors with the datasets from four GBM subgroups (CGAT, 2008; Verhaak et al., 2010). Based on GSEA, the transcriptome profile of Ctrl-T tumors was a significant match to the signature of the proneural subtype of GBM (Figure 7E). In contrast, signature gene sets associated with classical GBM, but not mesenchymal or neural subtypes, were significantly enriched in the gene expression profile of Olig2cKO tumors (Figure 7E).

qRT-PCR analysis further confirmed that expression of a set of astroglia-associated classical signature genes such as Egfr, Tbx2, Gli2, and Jag1 was substantially increased in Olig2cKO tumors (Figure 7F), whereas expression of proneural signature genes was downregulated. Expression of mesenchymal and neural signature genes was not significantly altered by Olig2 deletion (Figure 7F). In addition, we found that Olig2 directly targets the regulatory regions of the classical signature genes including Egfr, Tbx2, and Gli2 (Figure 7G). In Olig2cKO tumors, the activating histone mark H3K27ac was significantly enriched in enhancer regions of these genes, which were correlated with up-regulation of their expression in the absence of Olig2 (Figure 7G). This suggests that Olig2 negatively regulates expression of these classical signature genes in glioma.

To determine whether Olig2 down-regulation at the cellular level results in a shift of gene expression patterns, we knocked down Olig2 with lenti-shRNA in Ctrl-T tumor cells in culture. qRT-PCR was used to examine expression of a set of proneural signature (e.g., Olig2, Dll3, Ascl1, Slc1a1 and Pdgfra) and classical signature (e.g., Egfr, Tbx2, Gli2, and Jag1) genes (Verhaak et al., 2010). Strikingly, we found that the inhibition of Olig2 expression reduced proneural gene expression and led to an increase of expression in the classical signature genes such as Egfr compared to levels in control shRNA-treated cells (Figure 7H).

TCGA gene expression datasets and immunohistochemistry indicated that the tumors from the proneural GBM cohort of patients exhibit a high level of OLIG2 expression accompanied by a low level of EGFR expression, whereas a higher EGFR level was observed in the classical subtype cohort than in the proneural GBM cohort (Figures 7I, 7J, and S5A–S5C). This is in agreement with the previous observation that approximately 97% of patients with classical GBMs have high-level EGFR expression or amplification (Verhaak et al., 2010). Consistently, the proportion of OLIG2+ or OLIG2+/PDGFRA+ cells in GBM tissues was higher in the proneural group than the classical group (Figures S5D–S5F). There is a positive correlation between OLIG2 and PDGFRA expression among GBMs with PDGFRA amplification in TCGA datasets (Figure S5G), however, the levels of OLIG2 and EGFR expression are uncorrelated in TCGA core GBM samples. Although EGFR expression in OLIG2+ cells appears variable, ranging from absent or low to high, we observed that the percentage of OLIG2+ cells expressing low or absent EGFR was substantially higher than that of those with high EGFR expression in a cohort of proneural tumors (Figure S5H), consistent with a role of OLIG2 in repression of EGFR expression.

To further correlate OLIG2 down-regulation to gene expression changes, we knocked down expression of OLIG2 by shRNA in two primary human proneural GBM cell lines, TS543 and TS667; both lines characterized by PDGFRA amplification (Ozawa et al., 2014). We found that OLIG2 depletion by shRNA knockdown resulted in a reduction in the rate of cell proliferation as assayed by BrdU incorporation (Figures 7K and 7L). Similarly, OLIG2 knockdown in an OLIG2-expressing glioma cell line SU-AO2 with IDH1 mutation (Venkatesh et al., 2015) also caused a decrease in cell proliferation (Figures S5I and S5J). Furthermore, expression of characteristic genes of classical GBM, including EGFR, were up-regulated, while PDGFRA expression was downregulated upon treatment of cells with the OLIG2-specific shRNA in TS543 and TS667 cells (Figure 7M). Collectively, these observations suggest that OLIG2 has a cell-autonomous role on the phenotype of tumor cells.

Inhibition of EGFR signaling blocks Olig2-deleted tumor cell proliferation

We observed that Egfr was significantly up-regulated in Olig2cKO tumors, and that Olig2 directly targets the regulatory elements of Egfr and represses Egfr expression (Figure 7G). These results point to EGFR signaling that sustains reduced tumorigenesis in the Olig2cKO mice. Immunostaining and western blot analysis indicated robust EGFR up-regulation in Olig2cKO tumors in contrast to weak EGFR expression in Ctrl-T tumors (Figures 8A–8D), while EGFR+ cells in Olig2cKO tumors were proliferative (Figure 8B). In Olig2cKO tumors, expression of the activated form EGFR, phospho-EGFR, was also up-regulated (Figures 8C and 8D). Similarly, downstream targets of EGFR signaling, p-Erk and p-Akt, were increased in Olig2-deleted tumors, whereas overall Erk and Akt levels were unchanged relative to those in Ctrl-T tumors (Figures S6A and S6B). Consistent with activation of EGFR signaling, expression of EGFR ligands including EGF, HB-EGF, and TGFα were detected in Olig2cKO tumor tissues and all human TCGA GBM tumor samples (Figures S6C and S6D).

Figure 8. Targeting EGFR signaling suppresses tumor cell proliferation in Olig2cKO mice.

Figure 8

Open in a new tab

(A) Representative immunostaining images of Olig2, PDGFRα, and EGFR in Ctrl-T and Olig2cKO tumors. Scale bar, 50 µm.

B) Representative immunostaining images for EGFR, Ki67, and Olig2 in Ctrl-T and Olig2cKO tumors. Scale bar, 20 µm.

(C) Representative western blots show expression of PDGFRα, p-PDGFRα, EGFR, p-EGFR in Ctrl-T and Olig2cKO tumors. α-Tubulin as a loading control.

(D) Quantification of relative expression of PDGFRα, p-PDGFRα, EGFR, and p-EGFR in Olig2cKO and Ctrl-T tumors.

(E) Representative images showing the expression of PDGFRα (red) and Olig2 (green) in Ctrl-T and Olig2cKO tumors. Scale bar, 10 µm.

(F) The quantification of PDGFRα+ cells in Ctrl-T and Olig2cKO tumors per unit area (0.04 mm2).

(G) Representative images of BrdU (red) immunostaining of Ctrl-T and Olig2cKO tumor cells treated with gefitinib or vehicle (DMSO); Olig2 and DAPI staining was shown in green and blue, respectively. Scale bar, 30 µm.

(H) Percentages of BrdU+ cells in Ctrl-T and Olig2cKO tumor cells treated with gefitinib or DMSO. Data represent the means ± SEM from three independent experiments (** p < 0.01; *** p < 0.001; ANOVA with Newman–Keuls multiple comparison test).

(I) Quantification of cells using the CellTiter Glo viability assay from Ctrl-T and Olig2cKO tumor cells at indicated times after treatment. Plotted are cell numbers in gefitinib-treated samples divided by cell numbers from DMSO-treated samples.

(J) Cleaved Caspase 3 staining (green) in Ctrl-T and Olig2cKO tumor cells after DMSO or gefitinib treatment; Olig2 staining shown in red; DAPI in blue. Scale bar, 30 µm.

(K) Percentage of cleaved Caspase 3+ cells in Ctrl-T and Olig2cKO tumor cells treated with gefitinib or DMSO.

(L, M) Relative bioluminescence signals in (L) Ctrl-T mice at dpi 15 and dpi 25 and (M) Olig2cKO mice at dpi 30 and dpi 40 treated with DMSO (black) or gefitinib (gray) delivered by osmotic minipumps (n ≥ 6 mice for each group).

(N) Representative in vivo bioluminescence images of Ctrl-T and Olig2cKO mice treated with DMSO or gefitinib at indicated time-points post tumor induction.

(O) Kaplan-Meier survival analysis of Ctrl-T mice treated with DMSO (brown) or gefitinib (orange), and Olig2cKO mice treated with DMSO (blue) or gefitinib (green) delivered by osmotic minipumps. p < 0.05 with the log-rank test between DMSO or gefitinib treated Olig2cKO mice. n ≥ 5 mice for each group.

Data are presented as means ± SEM from at least three independent tumor tissue samples per group in D, F, I, K-M (* p < 0.05; ** p < 0.01; Student’s t test). See also Figure S6.

In contrast to EGFR, levels of PDGFRα and phospho-PDGFRα were substantially downregulated in Olig2cKO tumors (Figures 8C and 8D). The number of PDGFRα-expressing cells was decreased in Olig2cKO relative to Ctrl-T tumor tissues (Figure8E and 8F). In addition, we observed that Olig2 targets directly the enhancer regions of the Pdgfra gene locus in glioma (Figure S6E). Collectively, these observations suggest that Olig2 negatively regulates EGFR expression while activating PDGFRα expression in glioma.

Given that the EGFR pathway is activated in the Olig2-deleted tumors, we tested whether tumor cells with Olig2 deletion were sensitive to inhibition of EGFR signaling by treating tumor cells derived from Ctrl-T and Olig2cKO tumors with gefitinib, an EGFR inhibitor (Cataldo et al., 2011). Gefitinib treatment exhibited a stronger inhibitory effect on the proliferation of Olig2cKO tumor cells than Ctrl-T tumor cells as assayed by BrdU incorporation (Figures 8G and 8H). Similar results were also observed upon treatment of cells with erlotinib, another EGFR inhibitor (Cataldo et al., 2011) (data not shown). The number of viable cells detected by CellTiter Glo assay (Noah et al., 2007) was reduced significantly in Olig2-deleted cells after treatment with gefitinib, while Olig2+ Ctrl-T tumor cells were less sensitive to gefitinib treatment (Figure 8I). Furthermore, the number of cells undergoing apoptosis as marked by cleaved, active Caspase 3 was significantly higher in Olig2cKO cells in response to treatment than in Ctrl-T tumor cells (Figures 8J and 8K), where the majority of cleaved Caspase 3+ cells did not express Olig2 (Figures S6F and S6G). These data indicated that the growth of Olig2cKO cells is more sensitive to the EGFR inhibition than that of Olig2-expressing Ctrl-T tumor cells. The loss of Olig2 resulted in substantial reduction of PDGFRα but not PDGFRβ, which exhibited a low level in control tumors (Figures S6H–6J). We found that inhibition of PDGFR signaling with crenolanib had no significant effect on the growth of Olig2cKO tumor cells (Figures S6K and S6L). This suggests that in the absence of Olig2, tumor cells likely acquire the alternative EGFR signaling pathway to maintain the growth at a reduced rate.

Next, to determine the in vivo effects of gefitinib treatment on the tumor growth of Ctrl-T and Olig2cKO mice, we initiated drug treatment 5 days after PDGFB-Cre retrovirus induction of tumorigenesis. Ctrl-T mice that received gefitinib exhibited an insignificant reduction in the tumor size compared to mice that received vehicle as assayed bioluminescent intensity (Figures 8L and 8N). In contrast, Olig2cKO mice treated with gefitinib showed a significant reduction in bioluminescence over the course of treatment compared with mice treated with DMSO, indicating that gefitinib treatment substantially delayed tumor growth (Figures 8M and 8N). Gefitinib treatment resulted in a significant increase in survival time of the Olig2cKO animals (Figure 8O), suggesting that Olig2 depletion sensitizes glioma cells to EGFR signaling inhibition and improved animal lifespans.

Discussion

In the present study, we demonstrate that mitotic Olig2-expressing cells are essential for tumor propagation in a murine model of proneural-subtype GBM, highlighting the dependency of brain tumorigenesis on Olig2+ mitotic cells. Our findings also indicate that Olig2 reciprocally regulates distinct growth receptor PDGFR and EGFR signaling pathways and maintains tumor phenotype identity. Integrative analyses of genome occupancy and transcriptome profiling reveal that Olig2 directly activates a specific subset of genes that promote cell proliferation and oncogenic processes in the neoplasm distinct from normal NPCs. Olig2 deletion leads to a significant delay in the onset and progression of glioma and causes a phenotype shift from an oligodendrocyte-lineage correlated proneural toward a astroglial gene expression pattern with activation of EGFR signaling. Our data indicate a critical role for Olig2 in regulating the growth rate and phenotypic plasticity of both mouse and human glioma cells. We further find that Olig2 deletion results in sensitization of glioma cells to EGFR inhibition. Thus, our findings suggest a rationale for blocking Olig2-expressing proneural glioma growth through inhibiting Olig2 activity, and proof of principle to targeting tumor-propagating cells for stratifying therapy among distinct subtypes of malignant gliomas.

A previous study showed that Olig1/2-null cells were unable to form tumors in an allograft model in the presence of p53 (Ligon et al., 2007). Tumor cells however in our mouse model, which harbors deletions of Trp53 and Pten, the most frequently mutated genes in GBM patients (Brennan et al., 2013; Verhaak et al., 2010), are able to continue to grow in the absence of Olig2, albeit at a slower rate, revealing a context-dependent function of Olig2 in brain tumorigenesis. Olig2 has been shown to repress cell cycle inhibitor p21WAF1/CIP1 (Ligon et al., 2007) on the wild-type p53 background, and the requirement of Olig2 for tumor growth appears to be p53 dependent (Mehta et al., 2011). In the Trp53/Pten mutant glioma model, however, Olig2 deletion leads to up-regulation of EGFR but not p21 (data not shown). This is in keeping with the lack of the reciprocal relationship between Olig2 and p21 expression in glioma cells carrying mutated or amplified EGFR (Mehta et al., 2011). Strikingly, our integrative genome occupancy and expression profiling analyses reveal that Olig2 directly activates regulators of cell cycle and oncogenic factors in tumors but not normal NPCs, suggesting that Olig2 acts as an accelerator of cell division during tumorigenesis. Our findings indicate that regulation of cell proliferation and oncogenic pathways by Olig2 is required for full tumorigenic potential and rapid cell expansion during tumor growth.

Olig2-deleted tumor cells exhibit increased adhesiveness, astrocytic phenotypes, and a defect in neurosphere formation, suggesting that Olig2 is required for the robust self-renewal of tumor propagating cells, consistent with the finding of Olig2 as a key transcription factor that drives GBM (Suva et al., 2014). A recent study indicates that self-renewal capacity, as measured by an in vitro sphere formation assay, does not determine the tumor growth potential of high-grade glioma (Barrett et al., 2012). Consistent with this, our data show that, despite vastly diminished sphere formation capacity, Olig2-deleted cells are able to establish and form glioma, albeit at a decelerated rate. Together, our data suggest that Olig2 promotes the growth of tumor cells through activation of cell-proliferation activating factors and maintenance of the proliferation potential of glioma cells.

Recent studies indicate that TNF-α/NF-κB signaling promotes mesenchymal differentiation of proneural GBM cells in vitro (Bhat et al., 2013). Additional NF1 loss in a proneural glioma model induces proneural to mesenchymal transition (Ozawa et al., 2014). The molecular determinants that steer proneural subtype differentiation other than the mesenchymal lineage remain elusive. Gene profiling analysis indicates that Olig2-deleted tumors exhibit an increased expression of the gene signature of astrocytic classical-like GBM. In addition, Olig2 knockdown in mouse tumor cells or human proneural GBM cells results in increased expression of a set of classical signature genes including EGFR. Thus, although we could not completely exclude the possibility of a change in tumor cell compositions or microenvironment, these observations suggest that the phenotypic shift of Olig2cKO tumors occurs, at least in part, through a cell-autonomous effect of Olig2 loss.

Olig2-expressing control tumor cells are highly enriched in oligodendrocyte lineage signatures, whereas the Olig2cKO tumor cells were characterized by up-regulated expression of astroglial signature genes. This is consistent with the role of Olig2 in suppressing the astroglial phenotype by inhibiting expression of GFAP and astrogenic factors such as NFIA (Cai et al., 2007; Glasgow et al., 2014). Our observation of an apparent shift toward “classical” from “proneural” expression patterns based on the TCGA classification of gliomas likely reflects a switch from an oligodendroglial to astroglial gene expression pattern in the absence of Olig2. It is worth noting that Olig2 deletion does not necessarily cause a GBM subtype conversion. Rather, Olig2 deletion leads to a transcriptional program shift with apparent up-regulation of astroglia-correlated signature genes, while downregulation of oligodendrocyte precursor-associated proneural gene expression in our animal model.

Olig2+ mitotic cell elimination virtually abrogates tumor growth in our animal model of proneural-subtype gliomas, suggesting mitotic Olig2-expressing cells as a seeding source for proneural glioma growth. Although the potential role of mitotic Olig2+ cells in other subtype tumors remains to be determined, targeted elimination of OLIG2+ mitotic cells will be beneficial for treatment of a set of malignant GBM, especially for the proneural tumors that are enriched in mitotic OLIG2+ propagating cells. Notably, IDH1 or IDH2-mutant gliomas are highly enriched for OLIG2. IDH1/2 mutations cause a global DNA hypermethylation phenotype referred to as G-CIMP (Glioma CpG Island Methylator Phenotype) and widespread gene silencing (Brennan et al., 2013). Compared with other OLIG2-expressing non-G-CIMP tumors, IDH-mutant tumors are less aggressive (Joseph et al., 2013), raising a possibility of a context-dependent role of OLIG2 in tumor growth. Nonetheless, OLIG2 knockdown in IDH1-mutant SU-AO2 glioma cells results in a reduction of cell proliferation, suggesting that OLIG2 may also act as an oncogenic factor for the cell growth in IDH-mutant glioma. Collectively, our findings suggest that Olig2 controls glioma phenotype plasticity and that targeted inhibition of Olig2 sensitizes tumor cells to EGFR inhibition. The profound impact of targeting Olig2 and EGFR signaling on glioma formation illuminates an alternative avenue to stratify GBM treatment with tailored pharmacological intervention, to inhibit the growth of these highly lethal malignant brain tumors.

EXPERIMENTAL PROCEDURES

Animals, tissue samples, and Immunohistochemistry

Mouse lines, breeding and tissue processing used in this study are described previously (Yue et al., 2006) and in Supplemental Information. All animal experiments were approved by and performed according to the guidelines of the Institutional Animal Care and Use Committee at the Cincinnati Children’s Hospital Medical Center, USA. Human PN-like subtype GBM tissues with PDGFRA-amplification and CL-like subtype GBM tissues with EGFR amplification, CDKN2A deletion and PTEN deletion were from the University of Cincinnati, the Cincinnati Children’s Hospital Medical Center and Sichuan University. All human glioma samples were obtained with consent as outlined by institutional review boards at the individual Universities. Tissue processing and immunostaining procedures and antibodies used are listed in the Supplemental Information.

Tumor Induction and Intracranial Transplantation

Gliomas induced by PDGFB-IRES-Cre retrovirus and bioluminescence imaging in the Pten fl/fl;Trp53fl/fl mice were described previously (Lei et al., 2011) and in the Supplemental Information.

Drug Infusion

Mice were anaesthetized five days after PDGFB-Cre retrovirus injection, and implanted with osmotic minipump (Model 2004; Alzet-Durect Corp.) and cannulas (Brain Infusion Kit 3; Alzet-Durect Corp.). The infusion cannula was implanted through the same needle tract used for virus injection by the stereotaxic apparatus. The osmotic minipumps were filled up with vehicle (DMSO) or gefitinib (80 mg/ml) (flow rate 0.25 µl/hr, resulting in delivery of 480 µg/d) and were connected to the infusion cannulas.

Cell Culture, RNA-seq, ChIP-seq, and Data Analysis

Tumor cell isolation, culture, and sphere formation assays were described in the Supplemental Information. RNA isolation, RNA-seq, ChIP-seq and data analyses were performed as described previously (Yu et al., 2013). Full details can be found in the Supplemental Information.

Statistical Analysis

Statistical analyses were done using Microsoft Excel or Prism GraphPad 6.00 with Student's two-tailed t test for comparing two sets of data. One-way analysis of variance analysis (ANOVA) with a Newman–Keuls multiple comparison test for post-hoc analysis. Two-sided log-rank test was used to determine statistical significance of Kaplan-Meier survival curves. p < 0.05 is considered to be statistically significant (Supplemental Information).

Supplementary Material

Significance.

Distinct molecular characteristics in gliomas represent a significant hurdle for effective therapy. Here we uncover that a population of mitotic Olig2+ cells is critical for initiation and progression of mouse primary gliomas. Intriguingly, deletion of Olig2 results in tumors that exhibit persistent but decelerated growth. We find that Olig2 directly activates the transcriptional program for oncogenic growth. Olig2 deletion engenders a tumor phenotype shift from a proneural to an astrocytic gene expression pattern, with PDGF receptor-alpha down-regulation and reciprocal EGFR signaling upregulation. Down-regulation of Olig2 sensitizes glioma cells to EGFR inhibition. Our observations reveal that Olig2 is a molecular arbiter of tumor phenotype plasticity that may underlie drug resistance, suggesting new strategies for enhancing EGFR drug sensitivity in glioma treatment.

Highlights.

  1. Elimination of mitotic Olig2+ cells inhibits glioma initiation and progression

  2. Olig2 loss reduces glioma growth and causes proneural-to-astrocytic phenotype shift

  3. Olig2 deletion causes PDGFR down-regulation and reciprocal EGFR up-regulation

  4. Inactivation of Olig2 potentiates sensitization of glioma cells to EGFR inhibition

Acknowledgments

The authors would like to thank Xianyao Zhou, Zhixing Ma, Jason Lu, Xiaoting Zhu, and Peter Sims for technical support. We thank Dr. Michelle Monje Deisseroth for providing SU-AO2 tumor cell lines, and Drs. Yi Zheng, Charles Stiles, Susan Wells and Edward Hurlock for critical comments. This study was funded in part by grants from the National Institutes of Health (R01 NS078092 and R01 NS075243) to QRL.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCESSION NUMBERS

All the RNA-seq and ChIP-seq datasets are deposited in the NCBI Gene Expression Omnibus (GEO) GSE71493.

SUPPLEMENTAL INFORMATION

Supplemental Information includes Supplemental Experimental Procedures and six figures can be found with this article online.

AUTHOR CONTRIBUTIONS

Q.R.L. and F.L. designed the experiments, analyzed the data and wrote the manuscript with input from all authors. F.L., Y.C., C.Z., H.W., D.H., L.X., J.W., X.H., Y.D., E.E.L., X.L. and R.V. carried out the in vitro, in vivo, gene profiling or in silico analyses. H.B., R.D., M.F., A.O.S., D.B., M.X., J.B.R. E.M.B. provided samples and inputs. P.C. provided PDGF retroviral vectors and data interpretation. E.C.H. provided tumor cell lines and input. Q.R.L. supervised the project.

References

  1. Barrett LE, Granot Z, Coker C, Iavarone A, Hambardzumyan D, Holland EC, Nam HS, Benezra R. Self-renewal does not predict tumor growth potential in mouse models of high-grade glioma. Cancer Cell. 2012;21:11–24. doi: 10.1016/j.ccr.2011.11.025. [DOI] [PubMed] [Google Scholar]
  2. Bhat KP, Balasubramaniyan V, Vaillant B, Ezhilarasan R, Hummelink K, Hollingsworth F, Wani K, Heathcock L, James JD, Goodman LD, et al. Mesenchymal differentiation mediated by NF-kappaB promotes radiation resistance in glioblastoma. Cancer Cell. 2013;24:331–346. doi: 10.1016/j.ccr.2013.08.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Brennan CW, Verhaak RG, McKenna A, Campos B, Noushmehr H, Salama SR, Zheng S, Chakravarty D, Sanborn JZ, Berman SH, et al. The somatic genomic landscape of glioblastoma. Cell. 2013;155:462–477. doi: 10.1016/j.cell.2013.09.034. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Cai J, Chen Y, Cai WH, Hurlock EC, Wu H, Kernie SG, Parada LF, Lu QR. A crucial role for Olig2 in white matter astrocyte development. Development. 2007;134:1887–1899. doi: 10.1242/dev.02847. [DOI] [PubMed] [Google Scholar]
  5. Carro MS, Lim WK, Alvarez MJ, Bollo RJ, Zhao X, Snyder EY, Sulman EP, Anne SL, Doetsch F, Colman H, et al. The transcriptional network for mesenchymal transformation of brain tumours. Nature. 2010;463:318–325. doi: 10.1038/nature08712. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cataldo VD, Gibbons DL, Perez-Soler R, Quintas-Cardama A. Treatment of non-small-cell lung cancer with erlotinib or gefitinib. N Engl J Med. 2011;364:947–955. doi: 10.1056/NEJMct0807960. [DOI] [PubMed] [Google Scholar]
  7. CGAT. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature. 2008;455:1061–1068. doi: 10.1038/nature07385. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Chen J, Li Y, Yu TS, McKay RM, Burns DK, Kernie SG, Parada LF. A restricted cell population propagates glioblastoma growth after chemotherapy. Nature. 2012;488:522–526. doi: 10.1038/nature11287. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Friedmann-Morvinski D, Bushong EA, Ke E, Soda Y, Marumoto T, Singer O, Ellisman MH, Verma IM. Dedifferentiation of neurons and astrocytes by oncogenes can induce gliomas in mice. Science. 2012;338:1080–1084. doi: 10.1126/science.1226929. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Glasgow SM, Zhu W, Stolt CC, Huang TW, Chen F, LoTurco JJ, Neul JL, Wegner M, Mohila C, Deneen B. Mutual antagonism between Sox10 and NFIA regulates diversification of glial lineages and glioma subtypes. Nat Neurosci. 2014;17:1322–1329. doi: 10.1038/nn.3790. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Hayden EC. Genomics boosts brain-cancer work. Nature. 2010;463:278. doi: 10.1038/463278a. [DOI] [PubMed] [Google Scholar]
  12. Jansen M, Yip S, Louis DN. Molecular pathology in adult gliomas: diagnostic, prognostic, and predictive markers. Lancet Neurol. 2010;9:717–726. doi: 10.1016/S1474-4422(10)70105-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Johnson BE, Mazor T, Hong C, Barnes M, Aihara K, McLean CY, Fouse SD, Yamamoto S, Ueda H, Tatsuno K, et al. Mutational analysis reveals the origin and therapy-driven evolution of recurrent glioma. Science. 2014;343:189–193. doi: 10.1126/science.1239947. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Joseph NM, Phillips J, Dahiya S, M MF, Tihan T, Brat DJ, Perry A. Diagnostic implications of IDH1-R132H and OLIG2 expression patterns in rare and challenging glioblastoma variants. Mod Pathol. 2013;26:315–326. doi: 10.1038/modpathol.2012.173. [DOI] [PubMed] [Google Scholar]
  15. Lei L, Sonabend AM, Guarnieri P, Soderquist C, Ludwig T, Rosenfeld S, Bruce JN, Canoll P. Glioblastoma models reveal the connection between adult glial progenitors and the proneural phenotype. PLoS One. 2011;6:e20041. doi: 10.1371/journal.pone.0020041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Ligon KL, Alberta JA, Kho AT, Weiss J, Kwaan MR, Nutt CL, Louis DN, Stiles CD, Rowitch DH. The oligodendroglial lineage marker OLIG2 is universally expressed in diffuse gliomas. J Neuropathol Exp Neurol. 2004;63:499–509. doi: 10.1093/jnen/63.5.499. [DOI] [PubMed] [Google Scholar]
  17. Ligon KL, Huillard E, Mehta S, Kesari S, Liu H, Alberta JA, Bachoo RM, Kane M, Louis DN, Depinho RA, et al. Olig2-regulated lineage-restricted pathway controls replication competence in neural stem cells and malignant glioma. Neuron. 2007;53:503–517. doi: 10.1016/j.neuron.2007.01.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Liu C, Sage JC, Miller MR, Verhaak RG, Hippenmeyer S, Vogel H, Foreman O, Bronson RT, Nishiyama A, Luo L, et al. Mosaic analysis with double markers reveals tumor cell of origin in glioma. Cell. 2011;146:209–221. doi: 10.1016/j.cell.2011.06.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Lu QR, Sun T, Zhu Z, Ma N, Garcia M, Stiles CD, Rowitch DH. Common developmental requirement for Olig function indicates a motor neuron/oligodendrocyte connection. Cell. 2002;109:75–86. doi: 10.1016/s0092-8674(02)00678-5. [DOI] [PubMed] [Google Scholar]
  20. Meacham CE, Morrison SJ. Tumour heterogeneity and cancer cell plasticity. Nature. 2013;501:328–337. doi: 10.1038/nature12624. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Mehta S, Huillard E, Kesari S, Maire CL, Golebiowski D, Harrington EP, Alberta JA, Kane MF, Theisen M, Ligon KL, et al. The central nervous system-restricted transcription factor Olig2 opposes p53 responses to genotoxic damage in neural progenitors and malignant glioma. Cancer Cell. 2011;19:359–371. doi: 10.1016/j.ccr.2011.01.035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Meijer DH, Sun Y, Liu T, Kane MF, Alberta JA, Adelmant G, Kupp R, Marto JA, Rowitch DH, Nakatani Y, et al. An amino terminal phosphorylation motif regulates intranuclear compartmentalization of Olig2 in neural progenitor cells. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2014;34:8507–8518. doi: 10.1523/JNEUROSCI.0309-14.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Nie Z, Hu G, Wei G, Cui K, Yamane A, Resch W, Wang R, Green DR, Tessarollo L, Casellas R, et al. c-Myc is a universal amplifier of expressed genes in lymphocytes and embryonic stem cells. Cell. 2012;151:68–79. doi: 10.1016/j.cell.2012.08.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Noah JW, Severson W, Noah DL, Rasmussen L, White EL, Jonsson CB. A cell-based luminescence assay is effective for high-throughput screening of potential influenza antivirals. Antiviral Res. 2007;73:50–59. doi: 10.1016/j.antiviral.2006.07.006. [DOI] [PubMed] [Google Scholar]
  25. Ohgaki H, Kleihues P. Epidemiology and etiology of gliomas. Acta neuropathologica. 2005;109:93–108. doi: 10.1007/s00401-005-0991-y. [DOI] [PubMed] [Google Scholar]
  26. Ozawa T, Riester M, Cheng YK, Huse JT, Squatrito M, Helmy K, Charles N, Michor F, Holland EC. Most human non-GCIMP glioblastoma subtypes evolve from a common proneural-like precursor glioma. Cancer Cell. 2014;26:288–300. doi: 10.1016/j.ccr.2014.06.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Persson AI, Petritsch C, Swartling FJ, Itsara M, Sim FJ, Auvergne R, Goldenberg DD, Vandenberg SR, Nguyen KN, Yakovenko S, et al. Non-stem cell origin for oligodendroglioma. Cancer Cell. 2010;18:669–682. doi: 10.1016/j.ccr.2010.10.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Schonberg DL, Lubelski D, Miller TE, Rich JN. Brain tumor stem cells: Molecular characteristics and their impact on therapy. Mol Aspects Med. 2014;39:82–101. doi: 10.1016/j.mam.2013.06.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Sonabend AM, Bansal M, Guarnieri P, Lei L, Amendolara B, Soderquist C, Leung R, Yun J, Kennedy B, Sisti J, et al. The transcriptional regulatory network of proneural glioma determines the genetic alterations selected during tumor progression. Cancer Res. 2014;74:1440–1451. doi: 10.1158/0008-5472.CAN-13-2150. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Paulovich A, Pomeroy SL, Golub TR, Lander ES, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 2005;102:15545–15550. doi: 10.1073/pnas.0506580102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Suva ML, Rheinbay E, Gillespie SM, Patel AP, Wakimoto H, Rabkin SD, Riggi N, Chi AS, Cahill DP, Nahed BV, et al. Reconstructing and reprogramming the tumor-propagating potential of glioblastoma stem-like cells. Cell. 2014;157:580–594. doi: 10.1016/j.cell.2014.02.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Venkatesh HS, Johung TB, Caretti V, Noll A, Tang Y, Nagaraja S, Gibson EM, Mount CW, Polepalli J, Mitra SS, et al. Neuronal Activity Promotes Glioma Growth through Neuroligin-3 Secretion. Cell. 2015;161:803–816. doi: 10.1016/j.cell.2015.04.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Verhaak RG, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, Miller CR, Ding L, Golub T, Mesirov JP, 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. doi: 10.1016/j.ccr.2009.12.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Wang J, Wang H, Li Z, Wu Q, Lathia JD, McLendon RE, Hjelmeland AB, Rich JN. c-Myc is required for maintenance of glioma cancer stem cells. PLoS One. 2008;3:e3769. doi: 10.1371/journal.pone.0003769. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Yu Y, Chen Y, Kim B, Wang H, Zhao C, He X, Liu L, Liu W, Wu LM, Mao M, et al. Olig2 targets chromatin remodelers to enhancers to initiate oligodendrocyte differentiation. Cell. 2013;152:248–261. doi: 10.1016/j.cell.2012.12.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Yue T, Xian K, Hurlock E, Xin M, Kernie SG, Parada LF, Lu QR. A critical role for dorsal progenitors in cortical myelination. J Neurosci. 2006;26:1275–1280. doi: 10.1523/JNEUROSCI.4717-05.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Zhu X, Zuo H, Maher BJ, Serwanski DR, LoTurco JJ, Lu QR, Nishiyama A. Olig2-dependent developmental fate switch of NG2 cells. Development. 2012;139:2299–2307. doi: 10.1242/dev.078873. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

NIHMS774896-supplement.pdf (7.7MB, pdf)

RESOURCES