Olig2-regulated lineage-restricted pathway controls replication competence in neural stem cells and malignant glioma

Abstract

Recent studies have identified stem cells in brain cancer. However, their relationship to normal CNS progenitors, including dependence on common lineage-restricted pathways, is unclear. We observe expression of the CNS-restricted transcription factor, OLIG2, in human glioma stem and progenitor cells reminiscent of type C transit-amplifying cells in germinal zones of the adult brain. Olig2 function is required for proliferation of neural progenitors and for glioma formation in a genetically relevant murine model. Moreover, we show p21(WAF1/CIP1), a tumor suppressor and inhibitor of stem cell proliferation, is directly repressed by OLIG2 in neural progenitors and gliomas. Our findings identify an Olig2-regulated lineage-restricted pathway critical for proliferation of normal and tumorigenic CNS stem cells.

Figure 1. Human glioma progenitor cells and glioma stem cells express OLIG2

(A) Immunohistochemistry for OLIG2 (red) and the proliferation marker, Ki67 (green) in a human glioblastoma. Arrows denote co-localization of proteins (yellow) in glioma progenitor cells. (B) Scatter plot showing flow cytometric analysis of Ki67 and OLIG2 in a human glioblastoma sample. (C) Summary of flow cytometry results characterizing glioma progenitor cell populations in human glioblastomas (n=11). (D) Scatter plot showing flow cytometric analysis of CD133 and Olig2 in human glioblastoma sample. (E) Summary of flow cytometry results characterizing CD133+ putative glioma stem cell populations in fresh human GBM with respect to OLIG2 (n=11) and OLIG1 (n=4) expression. All p values < 0.001. O1=OLIG1, O2=OLIG2. Error bars indicate s.e.m. (F) Summary of the data shown in panels C and E.

Figure 2. Olig2 is required for glioma formation from neural stem cells

(A) Strategy for murine glioma model utilizing genetically defined neural stem cells as the cell of origin. 2 x 10⁵ Ink4a/Arf^−/−EGFRvIII neurosphere cells of the indicated genotypes were transplanted orthotopically into the striatum of SCID mice and animals monitored for symptoms of glioma formation. (B) Kaplan-Meier survival analysis. SCID mice injected with Olig1/2 ^+/− or Olig1^−/− Ink4a/Arf^−/−EGFRvIII neurospheres die from tumors with short latency (median survival=69 days and 81 days, respectively) relative to mice injected with Olig1/2-null Ink4a/Arf^−/−EGFRvIII neurospheres (96% survival at 273 days). Censored animals in the Olig1/2 ^−/− group (up ticks) indicate non-cancer deaths. p<0.01 for Olig1/2^−/− versus Olig1^−/− or Olig1/2^+/−. (C) Kaplan-Meier survival analysis showing restoration of tumor phenotype in Olig deficient mice by Olig2 encoding retrovirus (orange) (median TFS = 132 days), but not control GFP virus (purple) (100% survival at 136 days), p<0.01.

Figure 3. Olig2 is required for glioma growth but not neurosphere engraftment in the brain

(A, D) H&E staining and hEGFR IHC on coronal section of SCID mouse brain 100 days after injection with Olig1/2^+/− Ink4a/Arf^−/−EGFRvIII neurospheres demonstrating growth of malignant glioma. (B, E) Tumor histologically resembles human anaplastic astrocytoma W.H.O. grade III and uniformly expresses hEGFR. (C,F) Single cell infiltration of brain parenchyma is seen by H&E and hEGFR IHC. For additional characterization see Supplemental Figure 1. (G, J) Age matched brain injected with Olig1/2^−/− TNS. Box indicates injection site. (H) No tumor formation is detected by H&E staining at site of injection (arrow). (K) Rare surviving hEGFR+ Olig1/2−/− Ink4a/Arf^−/−EGFRvIII neurospheres have morphological characteristics of astrocytes (arrowhead) and neurons (arrow). (I, L) Subsequent infection of Olig1/2^−/− Ink4a/Arf^−/−EGFRvIII neurospheres with Olig2 encoding retrovirus rescues formation of gliomas histologically identical to those from Olig1/2^+/− Ink4a/Arf^−/−EGFRvIII neurospheres.

Figure 4. Olig2 regulates the growth and morphology of Ink4a/Arf^−/−EGFRvIII neurospheres

(A) The mean viable cell number of Olig1/2−/− Ink4a/Arf^−/−EGFRvIII neurospheres is significantly reduced compared to Olig1/2+/− controls (p<0.001). (B) Partial restoration of Olig1/2−/− Ink4a/Arf^−/−EGFRvIII neurosphere growth is seen following infection with Olig2 retrovirus but not with the GFP control (p<0.05). (C) Morphology of Olig1/2^+/− Ink4a/Arf^−/−EGFRvIII neurospheres (Tumorigenic Neurospheres, TNS) resembles that of normal neurospheres. (D) Olig1/2^−/− Ink4a/Arf^−/−EGFRvIII neurospheres exhibit marked adherence and migration on underlying substrate (arrow). (E) Infection of Olig1/2 −/− Ink4a/Arf^−/−EGFRvIII neurospheres with Olig2 retrovirus completely restores cells to the morphology of wild type Ink4a/Arf^−/−EGFRvIII neurospheres. All neurospheres were cultured under growth conditions in the presence of EGF.

Figure 5. Olig genes regulate the growth and morphology of normal neurospheres

(A–D) Olig2 is highly expressed in undifferentiated and proliferative neurosphere cells. (A) IHC showing widespread Olig2 expression (green) in a cultured neurosphere. (B) IHC for BrdU. (C) Merged image showing strong co-localization of Olig2 and BrdU. (D) Manual quantitation of IHC co-localization results for Olig1 and Olig2 expression (p<0.01). (E) The mean viable cell number at various passages for Olig1/2^−/− or Olig2^−/− neurospheres was significantly lower than wild type or heterozygous counterparts (p<0.001). (F) The percentage of proliferating BrdU+ cells is significantly lower in the Olig1/2^−/−cultures relative to wild-type (p<0.05). (G) Cell death measured by trypan blue staining is not significantly altered in the Olig1/2^−/− cultures compared to wild-type. (H) Typical rounded morphology of wild type neurospheres. (I, J) Olig1/2^−/− and Olig2^−/− neurospheres are reduced in numbers and size. Spheres also show extensive adherence and migration on the substrate (arrows). All neurospheres were derived from ganglionic eminences of wild type or Olig mutant 14.5 dpc embryos and were cultured in the presence of EGF.

Figure 6. A computational screen for direct genetic targets of Olig2

The screen is predicated on the functions of Olig2 protein as a transcriptional repressor (Mizuguchi et al., 2001; Novitch et al., 2001) that binds to canonical E-box regulatory elements of its target genes (Lee et al., 2005). Microarray expression profiling identified 882 probe sets that were derepressed (up-regulated) in Olig2-null neural stem cells relative to their wild type counterparts (see Supplementary Table 1). An additional 1021 probe sets corresponded to mRNAs that were down-regulated in Olig2-null cells relative to wild type (Supplementary Table 2). The list of derepressed probe sets was filtered through the MAPPER tool for binding site analysis to identified 167 genes with promoters enriched for E-box control elements. Analysis of this list of potential Olig2 targets revealed the presence of p21 and three additional p53-inducible genes (see Supplementary Table 3). A sampling of marker genes for stem cells, oligodendrocytes, astrocytes and neurons that are either derepressed (black font), repressed (red font) or unchanged (−) in Olig2-null cells relative to wild type is shown.

Figure 7. p21 is a direct target of Olig2 repression in neural stem/progenitor cells and human glioblastoma

(A) Diagram of P21 promoter region. Arrowheads indicate promoter regions analyzed by PCR in ChIP assays. E-boxes (red) and a p53 binding site (blue) are shown within the region that is bound by Olig2 (red box). (B) Quantitative PCR validates increased expression of p21 mRNA in Olig2-null (n=6) versus wild type (n=3) neurospheres (p<0.036). C) p21 luciferase reporter activity is repressed by Olig2 expression in a human GBM cell line (U87) while a dominant negative Olig2 (dnOlig2) defective in binding DNA yields increased expression (p values< 0.0001). (D) ChIP analysis using anti-Olig2 antibody to detect Olig2 binding to the p21 promoter in E14 rat cortical progenitor cells. Olig2 binding was only detected upon induction of Olig2 expression by FGF. Input is total DNA. (E) ChIP analysis of p21 promoter in a human glioma cell line transfected with a construct encoding a tamoxifen-activated Olig2 fusion protein. No binding to the related cell cycle inhibitors p27 or p57 is detected. Olig2 is detected using anti-ER antibody. (F) ChIP analysis demonstrating OLIG2 is bound to the p21 locus in vivo within freshly isolated human GBM specimens (n=3, T1–T3). Mock ChIP was conducted with normal rabbit IgG as control.