Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Feb 6.
Published in final edited form as: Cell Rep. 2013 May 30;3(6):2127–2141. doi: 10.1016/j.celrep.2013.04.035

Transcriptional distinctions between normal and glioma-derived A2B5+ progenitor cells identify a core set of genes dysregulated at all stages of gliomagenesis

Romane M Auvergne 1,2,*, Fraser J Sim 1,2,5,*, Su Wang 1,2, Devin Chandler-Militello 1,2, Jaclyn Burch 1,2, Yazan Al Fanek 1,2, Danielle Davis 1,2, Abdellatif Benraiss 1,2, Kevin Walter 3, Pragathi Achanta 6, Mahlon Johnson 4, Alfredo Quinones-Hinojosa 6, Sridaran Natesan 8, Heide Ford 7, Steven A Goldman 1,2
PMCID: PMC5293199  NIHMSID: NIHMS740097  PMID: 23727239

SUMMARY

Glial progenitor cells (GPCs) are a potential source of malignant gliomas. We used A2B5-based sorting to extract tumorigenic GPCs from human gliomas spanning WHO grades II-IV. mRNA profiling identified a cohort of genes that distinguished tumor-initiating progenitor cells (TPCs) from A2B5+ GPCs isolated from normal white matter. A core set of genes and pathways was substantially dysregulated in A2B5+ TPCs, that included the transcription factor SIX1, and its principal co-factors, EYA1 and DACH2. shRNAi silencing of SIX1 inhibited the expansion of glioma TPCs in vitro and in vivo, suggesting a critical and unrecognized role of the SIX1-EYA1-DACH2 system in glioma genesis or progression. By comparing the expression patterns of glioma TPCs to normal GPCs, we have identified a discrete set of pathways by which glial tumorigenesis may be better understood, and more specifically targeted.

INTRODUCTION

Gliomas are the most common primary intracranial neoplasms in humans, accounting for 80% of all malignant brain tumors. Current treatment strategies, including surgery, radiotherapy and chemotherapy, only modestly improve patient survival (Stupp et al., 2005). The limited efficacy of these approaches is a consequence of both the rapid invasion of brain tissue by glioma cells, and of the rapid appearance of both chemo- and radio-resistant lineages within treated tumors.

Gliomas may arise from transformed somatic stem and progenitor cells. Indeed, many types of primary CNS malignancies, including periventricular tumors (Sim et al., 2006), medulloblastomas, and gliomas (Hemmati et al., 2003; Ignatova et al., 2002) exhibit multipotentiality and self-renewal in vitro, suggesting their derivation from, or regeneration of, a stem cell phenotype. In glioma, a discrete cohort of CD133+ cells was first reported to be wholly responsible for the initiation of new gliomas in immunodeficient recipients (Bao et al., 2006; Singh et al., 2004). Yet a number of recent studies have enriched this picture, by observing that CD133-negative glioma cells are also tumorigenic, while not all CD133+ cells are so (Nishide et al., 2009); these observations suggest the co-existence of multiple tumor-initiating phenotypes in gliomas, most especially in high grade tumors.

The antigenic heterogeneity of tumor-initiating cells in gliomas may in part reflect the variety of potential cell types of origin of glioma. Several studies have reported that in rodents, gliomas may arise from neural stem cells of the ventricular subependyma (Alcantara Llaguno et al., 2009; Jackson et al., 2006). Yet while humans resemble rodents in harboring persistent subependymal neural progenitors (Pincus et al., 1998; Sanai et al., 2004), adult humans retain much larger populations of multipotential glial progenitor cells (GPCs) in the subcortical white matter, in numbers that dwarf those of the adult subependyma. They can be distinguished from other brain phenotypes by their expression of gangliosides recognized by MAb A2B5 (Nunes et al., 2003), as well as by NG2/CPSG4 and PDGFαR (Nishiyama et al., 2009; Sim et al., 2011). Interestingly, these prototypic GPC markers are overexpressed in glioma, (Ogden et al., 2008; Shih and Holland, 2006). Furthermore, in rodents, parenchymal GPCs can form tumors that assume the genomic and the histological profiles of human oligodendroglioma (Lindberg et al., 2009; Persson et al., 2010), as well as of glioblastoma (Assanah et al., 2006; Liu et al., 2011). Together, these data suggested that a proportion of human gliomas may arise from committed GPCs. On that basis, we asked whether a subtractive genomics strategy, comparing normal adult human GPCs to their identically-sorted counterparts derived from adult gliomas, would identify pathways specifically associated with human glial oncogenesis.

RESULTS

A2B5+ cells are abundant in glioma and manifest a stem cell phenotype in vitro

To assess the incidence of A2B5+ cells in adult gliomas, we used both immunomagnetic sorting (MACS) and flow cytometry. By MACS, A2B5+ cells were significantly more abundant in gliomas (14.4±1.3%, n=29) than in non-cancerous adult white matter (WM) or cortex (3.6±0.3%; n=54) (Figure 1A; Table S1). Flow cytometry, with less false negatives than MACS, confirmed that A2B5+ cells were significantly more abundant in tumors (28.8±3.2%; n=35) than in non-neoplastic WM (2.9±0.4%; n=3) (Figure 1B; Table S1). Among these A2B5+ glioma cells, while only a small fraction co-expressed CD133 (4.6±1.5%), almost half (45±5.4%) co-expressed the oligodendroglial protein OLIG2 (Ligon et al., 2007), and most (74±4.2%) also expressed the tumor marker SURVIVIN. The A2B5+ glioma cells were highly mitotic; 31±7.9% co-expressed Ki67 (Figures 1C;S1A). Of note, although CD133 was an uncommon phenotype in fresh tumors, most CD133+ cells (74±6.7%) co-expressed A2B5, as did most OLIG2+, Ki67+, and SURVIVIN+ cells (≥75%) (Figures 1D;S1A). To validate these cytometric data in situ, we identified glioma TPCs in sections of GBMs by their nuclear co-expression of OLIG2 and Ki67 (Ligon et al., 2007). While only a minority of OLIG2+ cells expressed Ki67 (34±5.2%; n=4), most Ki67+ cells expressed OLIG2 (78±3.8%) (Figure S1B-C). Similarly, while only 45±5.4% of GBM-derived A2B5+ cells expressed OLIG2 in vitro (Figure 1C), 82±8.1% of OLIG2+ cells were A2B5+ (Figures 1D;S1A). These data suggest that the A2B5 immunophenotype comprises a discrete fraction of glioma cells, which includes most CD133+ and OLIG2+/Ki67+/SURVIVIN+ cells.

Figure 1. In vitro and in vivo characterization of glioma-derived A2B5+ cells (A-B).

Figure 1

Open in a new tab

A2B5 based-magnetic cell sorting (MACS) (A) and flow cytometry analysis (B) of freshly dissociated gliomas and non-neoplastic white matter and cortex. Horizontal lines indicate the mean percentage of A2B5+ cells ± SEM. Triangles represent astrocytoma; squares represent oligodendroglioma; diamonds indicate oligo-astrocytoma (**p < 0.02, ***p < 0.001; Kruskal-Wallis test). NT: Non-tumor; LG: low grade; ANA: anaplastic; GBM: glioblastoma. (C-D) Dual flow cytometry (*) and immunocytochemical analysis of freshly isolated GBM cells stained with A2B5, CD133, Ki67, Olig2 and survivin. (E) Immunostaining of GBM-derived A2B5+ cells cultured under differentiating conditions for CNPase, β3- tubulin and GFAP markers (red). Nuclei were counterstained with DAPI (blue). (F) Coronal sections through a mouse brain transplanted with GBM-derived A2B5+ human cells, stained with the human-specific antigen antibody (red) (a-h); and (green) Ki67 (b), Sox2 (c), Nestin (f), Olig2 (d), NG2 (g), survivin (e), and GFAP (h). Cells were counterstained with DAPI (blue). (G-H) Hematoxylin and eosin stained sections of a patient’s tumor (H), and its derivative A2B5-sorted xenograft into an adult mouse (I).

A2B5-defined tumor progenitors are clonogenic, self-renewing and multipotential in vitro

We next asked if A2B5+ glioma cells were clonogenic in vitro. To that end, we isolated A2B5+ and A2B5− cells by FACS, and raised each fraction in low-density suspension culture, in serum-free media (Nunes et al., 2003). High-grade (HG) glioma-derived A2B5+ cells robustly generated neurospheres, with substantially greater efficiency than did their A2B5− homologues, as demonstrated by in vitro limiting dilution analysis (Figure S2A-D). To assess the lineage potential of A2B5-defined TPCs, isolated A2B5+ cells were next expanded as neurospheres, and then plated in 1% serum to promote their differentiation. Twelve days later, the resultant outgrowths were immunostained for the astrocytic marker GFAP, the neuronal markers βIII-tubulin and MAP2, and the early oligodendroglial markers olig2 and CNP. Each of these phenotypes was present within single spheres (Figure 1E), suggesting that A2B5+ TPCs, like normal GPCs (Nunes et al., 2003), retained multilineage competence in vitro.

Importantly, we found that A2B5 expression defined a preferentially expanding pool of glioma TPCs, and that its expression was sustained with propagation. When we established stable lines from WHO stage IV gliomas (n=5), and found that by the third passage, >98% of all cells were A2B5+ (Figure S2E). These lines have now been passaged for over a year, and each has retained A2B5 expression, with the concurrent expression of SOX2 (76±16.8%) and SURVIVIN (57±7.5%) (Figures S1A;S2F). Interestingly, while CD133 was only rarely expressed in freshly isolated glioma cells, most A2B5+ glioma cells developed CD133 expression with propagation (Figure S2G-H). Conversely, virtually all SOX2+, SURVIVIN+, OLIG2+, and Ki67+ cells in these cultures (>97% for each) - as well as most CD133+ cells (88±6.7%) - co-expressed A2B5 (Figure S1A). In contrast to the sustained self-renewal competence of A2B5+ TPCs sorted from HG gliomas and GBMs, those derived from low grade (LG) WHO II gliomas were only able to generate spheres at early passages. These cultures could not be maintained beyond 5 months, reflecting the limited self-renewal competence of low-grade glioma cells (Galli et al., 2004). These observations suggested that human GBM cells rapidly select for and acquire an A2B5 phenotype with repetitive passage in vitro, and that virtually all SOX2+ and CD133+ glial TPCs co-express A2B5+.

A2B5-defined tumor progenitors express telomerase as a function of grade

To assess the basis for the differential self-renewal competence of LG and HG glioma-derived A2B5+ cells, we asked whether these cells expressed telomerase enzymatic activity (Langford et al., 1995), and if so, at what stages of anaplastic progression. Using the telomeric repeat amplification protocol assay (TRAP), we detected a significant degree of telomerase reverse transcriptase activity in 5 of 6 GBM-derived A2B5+ cells. In contrast, no detectable telomerase activity was detected in either LG-derived A2B5+ cells, or those derived from normal adult white matter (Figure S3A). Together, these data suggest that telomerase induction is associated with malignant progression rather than with initial gliomagenesis, and provide a basis for the limited self-renewal competence of LG glioma-derived A2B5+ TPCs.

A2B5-sorted cells initiated new gliomas in immunodeficient recipients upon orthotopic graft

To determine if A2B5+ TPCs could independently initiate gliomas in naive hosts, freshly dissociated cells from 6 WHO stage II to IV gliomas were sorted using MACS or FACS into A2B5+ and A2B5− fractions and were then transplanted, over a range of 5,000 – 100,000 cells/graft, into the corpus callosa of immunodeficient mice. The resultant xenografts were analyzed 6 – 15 weeks later (Figure S4A). Among 18 mice implanted with A2B5+ cells, 17 developed invasive gliomas by 6 weeks, while among 16 mice transplanted with A2B5− cells from the same tumors, 13 did so. No significant differences were noted in the distribution, volume, or Ki67 index of tumors derived from MACS-sorted A2B5+ and A2B5− cells (Figure S4B-E). Interestingly, the 3 mice that failed to develop tumors from A2B5− grafts were those given the lowest dose, 5,000 cells, which was invariably tumorigenic when A2B5+ cells were used. Thus, both A2B5+ and A2B5− TPCs were tumorigenic upon xenograft, although in vivo limiting dilution analysis suggested preferential tumorigenesis by the A2B5+ fraction, as reported (Ogden et al., 2008; Tchoghandjian et al., 2009). The A2B5+ xenografts were both highly migratory (Figure 1Fa) and highly proliferative, with a Ki67 index of 27.1±2% (n=11) (Figure 1Fb). Like the tumors from which they derived, the grafts expressed survivin, sox2, nestin, olig2, NG2, and GFAP (Figure 1F c-h), and assumed the histological appearance of native GBMs (Figure 1G-H).

Expression profiling revealed an A2B5+ TPC gene expression signature

We next sought to identify the molecular concomitants to the transformation of A2B5+ TPCs, using Affymetrix U133+2 microarrays. A2B5+ tumor cells, derived from both LG (WHO grade II, n=10), and HG (WHO grade III-IV, n=10) glioma (Table S2) were compared to their A2B5+ counterparts derived from normal adult WM (n=4) and cortex (n=4). In addition, since microglia may be recognized by both the A2B5 and NG2 antibodies (Pouly et al., 1999), we also profiled CD11b+ human microglia sorted from adult epileptic resections (n=4), so as to establish a filter by which to exclude microglial transcripts from our TPC gene sets (Figure 2A). Principal component analysis (PCA) confirmed that A2B5+ cells isolated from both LG and HG glioma showed overall expression profiles that were readily distinguished from those of normal adult GPCs, unsorted tumor cells, and microglia (Figure 2B).

Figure 2. Differential gene expression profiles of A2B5+ glioma cells at all stages of gliomagenesis.

Figure 2

Open in a new tab

(A,C) Venn diagram (A) and Table (C) showing the number of genes up- or down-regulated (>3 fold change, 1% false discovery rate) in A2B5+ TPCs versus both normal adult A2B5+ GPCs and CD11b+ microglial cells (MG) using Affymetrix expression microarrays. (B) Principal Component Analysis (PCA). Green and purple, Low- and high-grade derived A2B5+ cells; red, non-tumor A2B5+ GPC; tangerine, unsorted normal cells; green cross, microglial cells.

To define those genes whose expression distinguishes A2B5+ TPCs (n=20) from non-neoplastic GPCs (n=8), we performed differential gene expression analysis using relatively stringent cut-offs (>3- fold change, 1% FDR) and identified a total of 355 dysregulated genes (226 up, 113 down) (Figure 2C, Tables S3-S4). To further identify grade-associated changes in A2B5 expression signature, we separately compared the profiles of A2B5+ cells derived from LG and HG gliomas to those of A2B5+ GPCs, as well as to one another, so as to identify those genes associated with anaplastic progression (Figure 2C, Tables S3-4). We then separately compared the expression patterns of oligodendroglioma (OLG) and astrocytoma (AST) A2B5+ cells (WHO grade II) to one another, as well as to normal A2B5+ GPCs (Figures 2C;S5, Tables S3-4) (Figure 2C), so as to distinguish those gene sets associated with early stages of glioma progression from those involved in fate determination.

Among the genes dysregulated in glioma-derived TPCs, we focused initially on those genes with the highest expression ratios in all A2B5+ tumor cells relative to normal GPCs Figure 3A and Table 1). Only 8 genes were found that were ≥10-fold over-expressed in TPCs at all levels of anaplastic progression (Figure 3A), among which CD24, GAP43, MMP3 and IGFBP3 have been previously associated with invasive glioma. In addition, this analysis revealed a set of genes not previously known to be involved in gliomagenesis, that included SIX1, EYA1, SATB2 and CSRP2. Interestingly, the transcription factor SIX1, and its co-activating binding partner EYA1 have been shown to participate in the oncogenesis of human mammary carcinoma cells (Christensen et al., 2008; Pandey et al., 2010). Similarly, SATB2 and CSRP2 have been related to disease progression in carcinoma (Midorikawa et al., 2002; Patani et al., 2009). Several other oncogenes were also highly over-expressed, though less than 10-fold; these included EGFR, MYC, and the inhibitor of differentiation proteins ID1 and ID4 (Table S3). In addition, a number of genes were down-regulated by A2B5+ TPCs at all stages of progression (n=113) (Table S4), 6 by >10-fold (Figure 3B). Several of these have been described as tumor suppressors, including MTUS1 (Di Benedetto et al., 2006) and SPOCK3/testican 3 (Earl et al., 2006). Quantitative- PCR confirmed the dysregulation of selected genes (Figure 3C and Table S9). By defining those gene sets dysregulated in A2B5+ TPCs relative to their normal adult homologues in both LG and HG gliomas, we identified a discrete cohort of genes associated with both the initial appearance and anaplastic progression of glioma.

Figure 3. Specific genes and pathways dysregulated in A2B5+ glioma cells at all stages of gliomagenesis.

Figure 3

Open in a new tab

(A-B) Heat-map representation of the top up-regulated (A) and down-regulated (B) genes (>10 FC) in A2B5+ TPCs, as assessed by microarray analysis and compared to normal A2B5+ GPCs derived from adult human white matter (WM), cortex (CTX), or CD11b+ microglia (MG). (C) Real time-PCR validation of selected genes deregulated in A2B5+ TPCs relative to normal A2B5+ GPCs. Real-time PCR was performed using a 96-gene Taqman Low Density Array (TLDA), or individual pre-validated Taqman assays (as indicated by *). Gene expression was normalized to GADPH. FC: Fold change. (D-E) Heat-map representation of Gene-Ontology and KEGG-based pathways analysis of the TGF-β (D) and the Wnt (E) pathways in A2B5+ TPCs relative to their non-neoplastic counterparts isolated from the adult human WM and CX, and CD11b+ MG. MG: microglia; UNS: unsorted cells; CTX: A2B5+ cells from normal cortex; WM: A2B5+ cells from normal white matter; AST: Astrocytoma; AAST: anaplastic astrocytoma; OLG: Oligodendroglioma; GBM: Glioblastoma; GBM sc: small cell GBM; GSC, gliosarcoma.

Table 1. Significantly dysregulated genes in glioma-derived A2B5+ cells relative to normal A2B5+ GPCs.

Significantly dysregulated genes (>3-fold change, 5% FDR) were annotated into functionally relevant categories. The top five from each category are shown in the table.

Category Gene symbol Description Fold Change q-value
Transcription SIX1 SIX homeobox 1 16.41 7.23E-10
Factors SATB2 SATB homeobox 2 10.41 7.03E-12
FOXD1 forkhead box D1 8.27 1.41E-08
LOC100287917 hypothetical protein LOC100287917 8.22 6.57E-07
MYC v-myc / c-myc 8.05 1.47E-08
SMARCA2 SWI/SNF related, matrix associated, actin dependent regulator of chromatin −6.40 9.13E-09
ST18 suppression of tumorigenicity 18 (breast carcinoma) (zinc finger protein) −3.06 6.32E-03
Ligands IL33 interleukin 33 6.91 6.71E-06
TNC tenascin C 6.24 3.28E-05
INHbA Activin beta A 5.98 2.38E-06
STC2 stanniocalcin 2 5.97 7.82E-09
MIF macrophage migration inhibitory factor (glycosylation-inhibiting factor) 5.68 2.39E-08
GREM1 Gremlin −6.26 1.83E-04
GPNMB glycoprotein (transmembrane) nmb −6.00 2.23E-07
Receptors EGFR epidermal growth factor receptor 9.04 3.86E-07
IL13RA2 interleukin 13 receptor, alpha 2 8.04 4.38E-04
CNR1 cannabinoid receptor 1 (brain) 5.21 2.66E-07
F2R coagulation factor II (thrombin) receptor 4.59 2.84E-07
ADRA2A adrenergic, alpha-2A-, receptor 3.97 2.02E-04
GPR37 G protein-coupled receptor 37 (endothelin receptor type B-like) −7.87 1.43E-05
TEK TIE2 −6.28 6.31E-06
GRM3 glutamate receptor, metabotropic 3 −5.63 9.65E-07
LRP2 low density lipoprotein-related protein 2 −5.48 1.21E-04
P2RY12 purinergic receptor P2Y −3.80 4.50E-03
Other/Novel CD24 CD24 antigen (small cell lung carcinoma cluster 4 antigen) 34.53 1.33E-11
LOC100288551 hypothetical protein LOC100288551 19.31 6.17E-11
GAP43 growth associated protein 43 15.00 8.99E-10
IGFBP3 insulin-like growth factor binding protein 3 11.77 2.41E-05
CSRP2 cysteine and glycine-rich protein 2 9.91 3.49E-09
ANKRD43 ankyrin repeat domain 43 −20.75 1.29E-10
C21orf131 chromosome 21 open reading frame 131 −16.10 5.39E-09
MOBP myelin-associated oligodendrocyte basic protein −14.74 3.29E-10
OPALIN oligodendrocytic myelin paranodal and inner loop protein −11.98 2.81E-09
MTUS1 microtubule associated tumor suppressor 1 −10.08 1.61E-11
Enzymes/ MMP3 matrix metalloproteinase 3 (stromelysin 1, progelatinase) 19.75 1.12E-06
catalytic EYA1 eyes absent homolog 1 (Drosophila) 12.67 3.82E-09
MGST1 microsomal glutathione S-transferase 1 9.48 4.83E-07
METTL7B methyltransferase like 7B 9.09 6.89E-07
EGFR epidermal growth factor receptor 9.04 3.86E-07
ENPP2 ectonucleotide pyrophosphatase/phosphodiesterase 2 (autotaxin) −9.02 2.07E-08
C5orf4 chromosome 5 open reading frame 4 −8.39 2.14E-07
PDK4 pyruvate dehydrogenase kinase, isozyme 4 −7.97 4.49E-08
ACACB acetyl-Coenzyme A carboxylase beta −6.99 1.75E-07
GATM glycine amidinotransferase (L-arginine:glycine amidinotransferase) −6.57 5.44E-07
ECM/ MMP3 matrix metalloproteinase 3 (stromelysin 1, progelatinase) 19.75 1.12E-06
Cell adhesion LAMB1 laminin, beta 1 8.24 2.32E-04
TNC tenascin C 6.24 3.28E-05
CCDC80 coiled-coil domain containing 80 5.67 1.35E-07
SPOCK3 sparc/osteonectin, cwcv and kazal-like domains proteoglycan (testican) 3 −18.27 1.42E-08
Open in a new tab

A2B5+ glioma cells over-expressed TGFβ, BMP and wnt pathway components

To identify those pathways most selectively dysregulated in A2B5+ glioma TPCs relative to their normal homologues, we applied a set of functional (GO, KEGG) and pathway (mSigDB, IPA) databases to the list of differentially expressed genes (by >3 FC, 1% FDR). Our analysis revealed a strong enrichment for genes associated with cancer, cell proliferation, cell migration and motility, in both LG- and HG-derived A2B5+ cells, suggesting that neoplastic A2B5+ cells have greater proliferative and migration competence than their homologues derived from normal brain (Table S5; Figure S6B). Among specific gene sets that were over-represented in A2B5+ TPCs, GO revealed a significant enrichment of genes involved in both wnt/β-catenin and BMP signaling, while KEGG analysis highlighted the TGFβ signaling pathway as the most significantly over-represented (Table S5). Ingenuity pathway analysis (IPA) confirmed the predominant association of both TGF-β pathway and wnt/β-catenin-associated genes with glial TPCs (Figures 3D-E and Tables S5 and S8). Similarly, gene set enrichment analysis (GSEA) revealed an over-representation in A2B5+ TPCs of MYC target genes (Table S5).

A2B5+ cells derived from low-grade glioma expressed a proneural signature

Having defined the genes differentially expressed in A2B5+ TPCs at all stages of progression, we next focused on those genes accompanying early tumorigenesis. The expression profiles of LG-derived A2B5+ cells were compared to those of their normal A2B5+ homologues, identifying a set of 161 differentially-expressed genes (Figure 2C). Among these were a small cohort of >10-fold over-expressed genes; these included tumor-associated transcripts such as CD24, EYA1, SIX1, but also neurogenesis-associated genes, such as NEUROD1, INA, SATB2, ELAVL2, all suggestive of proneural phenotype (Figure 4A; Table S6). In contrast, among those genes significantly down-regulated in LG A2B5+ cells; we identified several tumor suppressors not previously associated with gliomagenesis, including MTUS1, GPNMB, and RGN, as well as several markers of mature oligodendrocytes, including MOBP and OPALIN, and ASPA (Figure 4B; Table S6). In addition, GO- and KEGG-based analysis revealed a significant up-regulation of genes associated with the BMP and the TGFβ signaling pathways, while mSigDB similarly revealed enrichment of MYC-induced and associated genes (Table S6). IPA revealed an over-representation of genes related to wnt/β-catenin and AMPK signaling pathways (Tables S6 and S8).

Figure 4. Expression profiles of A2B5+ cells during early tumorigenesis and anaplastic progression (A-B).

Figure 4

Open in a new tab

Heat-maps representing the top 26 up-regulated (A) ( > 10 FC, 1% FDR) and 16 down-regulated genes (B) (> 5FC, 1% FDR), by microarray analysis of LG glioma-derived A2B5+ TPCs, relative to normal A2B5+ GPCs isolated from adult white matter (WM), cortex (CTX) or CD11b microglial cells (MG). (C-D) Heat-maps representing the top up-regulated (C) and down-regulated genes (D) (> 10 FC, 1% FDR), by microarray analysis of A2B5+ cells isolated from HG (WHO III-IV) tumors relative to their LG (WHO II) counterparts. FC: fold change; FDR: False Discovery Rate.

LG and HG A2B5+ TPCs exhibit expression profiles of known molecular subclasses of GBM

We next used gene set enrichment analysis (GSEA) to evaluate the relationships of the A2B5+ TPC signature to previously established tissue-based data sets describing the major molecular subclasses of human GBM; and GBM-derived CD133+ TPCs gene sets (Figure 5A). We first compared the profiles of neoplastic A2B5+ cells to unsorted non-tumor cells, and observed a preferential enrichment of the proneural signature (PN) in LG tumors while HG-derived A2B5+ TPCs were enriched for gene sets that typify the epithelial-mesenchymal transition (EMT); the proliferative (PROLIF) and the classical (CL) subtypes of GBM (Phillips et al., 2006; Verhaak et al., 2010) (Figures 5C;S7; S8 A-C). As such, LG A2B5+ TPCs were enriched in gene sets associated with relatively prolonged patient survival, while the outcomes associated with the gene sets of HG A2B5+ TPCs are less favorable (Freije et al., 2004). Importantly, both LG- and HG-derived A2B5+ cells exhibited a significant enrichment of the CD133-UP signature, suggesting significant molecular homology between these tumor-initiating phenotypes (Figure 5C).

Figure 5. Molecular homologies among A2B5+ TPCs, CD133+ TPCs, gliomas, hESCs and normal GPCs.

Figure 5

Open in a new tab

(A-H) Enrichment patterns of the A2B5+ glioma expression signature across previously established human glioma expression profiles, CD133+ TPCs, human embryonic stem cells (hESC), and fetal human brain CD140a+ glial progenitor cells (GPCs) using parametric gene set enrichment analysis (PGSEA). (A) List of gene sets used associated with molecular subclasses of human gliomas and CD133-defined TPCs. (B) Gene sets associated with hESCs, pluripotentiality-associated transcripts, and CD140a+ GPCs. (C-H) PGSEA analysis of neoplastic A2B5+ cells compared to normal unsorted cells (C,F), neoplastic A2B5− cells (D,G) and normal A2B5+ cells (E,H). To identify significant enrichment/depletion, a linear model approach was used and p-values for each comparison were corrected for multiple testing using false discovery rate (q-values). Significance indicated as: *q < 0.05, **q < 0.01, ***q < 0.001. LG: low grade; HG: High grade.

Interestingly, glioma-derived A2B5+ cells also exhibited enrichment of the PN signature relative to their A2B5− counterparts, especially so in LG tumors (Figures 5D;S8C). In contrast, the PN signature was depleted in neoplastic A2B5+ cells relative to non-neoplastic GPCs, whereas MES/EMT related gene sets were relatively enriched (Figures 5E;S7;S8B). Together, these results indicate that LG and HG glioma-derived A2B5+ cells respectively share features of PN and MES subclass GBMs, as well as of GBM-derived CD133+ TPCs, and that both may be distinguished from non-neoplastic GPCs by their selective over-representation of MES/EMT related genes, which becomes more pronounced with anaplastic progression.

Anaplastic progression of A2B5+ TPCs reflected an epithelial-to-mesenchymal transition

To more precisely define the grade-associated evolution in gene expression by A2B5+ TPCs, we next compared the expression profiles of A2B5+ cells derived from HG versus LG gliomas (>3 FC, 1% FDR), and identified a total of 683 dysregulated genes (Figure 2C; Table S7). Among these, 17 were over-expressed by >10-fold in HG-derived A2B5+ cells; not surprisingly, these transcripts were related to transformation, invasion, angiogenesis, and to a mesenchymal differentiation, (Figure 4C). GSEA confirmed that many of these progression-associated genes were associated with mesenchymal phenotype and the epithelial-to-mesenchymal transition (Figure S7B and Table S7).

Among those genes that were down-regulated in HG-derived A2B5+ TPCs relative to their LG counterparts (n=225) (Figure 4D, Table S7), we identified the tumor suppressors SSTR1 (Patel, 1999) and SPOCK3 (Earl et al., 2006); the SIX1 repressor DACH2 (Christensen et al., 2008), and SHISA2, a regulator of the wnt-β/catenin and FGF signaling pathways (Hedge and Mason, 2008). KEGG-based functional analysis and IPA confirmed the relative dysregulation in HG A2B5+ TPCs of wnt/β-catenin signaling, as well as of IGF1, p53, Notch, PI3/AKT, and JAK-STAT dependent pathways (Table S7). GSEA also demonstrated the differential over-representation of genes involved in cell motility, TGFβ signaling pathway, and MYC target genes (Table S7). Together, these data indicate that the anaplastic progression of the A2B5+ phenotype is associated with dysregulated gene expression within the wnt/β-catenin and TGFβ pathways, concurrent with an EMT transition, which is itself attended by both the increased expression of genes associated with motility and parenchymal invasion, and the down-regulation of a discrete set of known tumor suppressors.

Low- and high-grade derived A2B5+ cells share common gene expression profiles with human ES cells and GPCs

To further explore the precursor nature of A2B5+ glioma cells, we used GSEA to compare the gene expression patterns of A2B5+ TPCs to human embryonic stem cells (hESC) and CD140a+ (PDGFRα) GPCs, using published data sets (Figure 5B, F-H). Interestingly, the hESC signature was enriched in both LG and HG derived A2B5+ cells (Figure 5 F-H). In addition, when compared to unsorted non-tumor brain cells as well as to A2B5− glioma cells, both LG and HG-derived A2B5+ cells were strongly enriched for the CD140a+ signature (Figure 5F-G). Accordingly, both microarray and qPCR analysis of marker genes revealed the selective enrichment in A2B5+ TPCs of both GPC and NSC marker genes (Figure S9A-J). Interestingly, glioma-derived A2B5+ cells could be readily distinguished from their non-tumor homologues by the expression of the stem and mesenchymal marker CD44 (Figures S9D,E,H), reflecting the selective expression by A2B5+ glioma TPCs of the MES signature, relative to normal, non-neoplastic GPCs.

SIX1 inhibition prevented the growth, the proliferation and the survival of GBM-derived TPCs

The homeobox transcription factor SIX1 and its co-factor EYA1 were among the highest differentially expressed tumor transcripts in A2B5+ TPCs relative to their non-neoplastic homologues (Figure 3C; Tables 1;S9). qPCR confirmed that SIX1 mRNA was highly over-expressed by A2B5+ cells isolated from both LG (231±128 FC, n=4; p < 0.001) and HG (76±25 FC, n=8; p<0.001) gliomas relative to their non-tumor A2B5+ counterparts (n=4), as well as in GBM-derived TPC lines (88±51 FC, n=5; p=0.016) (Figures 3C;S10A). Western immunoblots similarly revealed the high level expression of Six1 protein by primary gliomas and glioma cell lines, and its absence from the adult human brain (Figure S10B).

The robust overexpression of SIX1 in A2B5-defined TPCs, paired with its essential absence from the normal adult brain, prompted us to examine its contribution to gliomagenesis. To determine whether glioma TPCs were dependent on Six1 signaling, we first tested the effects of lentiviral-induced knockdown (KD) of SIX1 (Figure S10C-D) on the growth of glioma TPCs - in vitro. Lentiviral shRNAi silencing of SIX1 (SIX1 KD) significantly reduced the number of GBM-derived TPCs relative to both scrambled (SCR) shRNAi-transduced and non-transduced control (CT) cells, 6 days post-transduction (P=0.0005) (Figures 6A-B). On that basis, we next investigated the effects of Six1-induced KD on cell proliferation, cell cycle progression and cell survival. We found that SIX1 KD significantly reduced the mitotic fraction of BrdU-incorporating TPCs (24.3±4.5%), relative to both SCR (32±3.8%) and CT cells (36.3±3.4%) (P=0.006) (Figure 6C). We next addressed the role of Six1 in cell cycle progression, by analyzing EdU incorporation in association with propidium iodide (PI) staining. TPCs subjected to SIX1 KD manifested fewer cells in S phase (5.2±1.1%) relative to SCR (8.8±1.5%) and CT cells (9±1.2%) (P=0.018) (Figure 6D). We next asked if SIX1 suppression might be associated with increased cell death, and found that SIX1 KD TPCs exhibited a significantly increased incidence of Annexin V-defined apoptotic death (33.9±7.8%), compared to both SCR (24.7±8.9%) and CT cells (20.3±7.2%) (P=0.004; all comparisons by repeated measures ANOVA) (Figure 6E).

Figure 6. Six1 knock-down impairs the growth, proliferation and survival of A2B5+ TPCs in vitro and in vivo.

Figure 6

Open in a new tab

Effects of Six1 silencing on the in vitro growth (A-B), proliferation (C-D), and survival (E), as well as the in vivo tumorigenicity (F), of A2B5+ TPCs at 6 days and 6 weeks, respectively, after transduction with either Six1-KD lentivirus, scrambled shRNAi lentivirus, or non transduced control. (A-B) Representative photomicrographs (A) and graph (B) illustrating glioma cell number after transduction of TPC glioma lines by SIX1 KD or control vectors. (C-D) Graphs illustrating the percentage of BrdU (C) immunolabeling; and EdU (D) incorporation in association with propidium iodide. (E) (E-F) Effect of SIX KD on both apoptotic cell death as determined by flow cytometric analysis of Annexin V (E), and on the number (F) of hNA+ cells following transplantation of 2 distinct lines of glioma-derived A2B5+ TPCs, each transduced with either SIX KD or control shRNAis. KD: Knock-down. Means ± SEM. P-values calculated using one-way ANOVA with repeated measures followed by Tukey post-hoc comparisons with *p < 0.05; ** p < 0.01; ***p < 0.001

On the basis of these in vitro data, we examined the effects of SIX1 inhibition on the tumorigenic competence of A2B5+ TPCs in vivo. TPC lines established from A2B5+ cells derived from an anaplastic oligodendroglioma and a GBM were transplanted into the brains of immunodeficient mice (6-8 × 104 cells/animal, n=3 animals/group), 6 days after lentiviral transduction. SIX1 silencing inhibited the tumorigenicity of glioma TPCs, as demonstrated by a significant decrease in the number of hNA+ cells at 6 weeks post-transplantation, relative to SCR and CT cells (P=0.01) (Figures 6F,S10E). These observations indicate that SIX1 KD potently inhibits the growth, proliferation and survival of A2B5-defined glioma TPCs, both in vitro and in vivo, and suggest that SIX1 plays a critical role in the initiation and/or the proliferative expansion of malignant glioma.

DISCUSSION

In this study, we identified a core set of genes and pathways that are dysregulated throughout the anaplastic progression of glioma. We did so by comparing the gene expression patterns of A2B5-sorted GPCs derived from the normal adult human brain, to the expression patterns of their presumed homologues derived from gliomas at varying stages of progression ranging from WHO II low-grade (LG) gliomas to grade IV glioblastoma (GBM). We first validated both the robust self-renewal competence of A2B5+ glioma TPCs, and their diffuse invasion and efficient gliomagenesis when transplanted to immunodeficient hosts. We next assessed those genes and pathways differentially expressed by A2B5+ TPCs relative to their normal homologues, and identified tumor specific A2B5 transcript signatures as a function of both tumor stage and phenotype. Both LG and high-grade (HG) glioma-derived A2B5 signatures shared salient features of the transcription patterns of known GBM subtypes, GBM-derived CD133+ TPCs, and human ES cells. In particular, we identified the TGF-β and the wnt/β-catenin signaling pathways, as well as MYC, as potently dysregulated in neoplastic A2B5+ cells at every stage of tumor progression. These observations were in line with previous reports (Ikushima et al., 2009; Pulvirenti et al., 2011; Wang et al., 2008) and suggest that TGF-β, wnt/β-catenin and MYC signals all contribute to the initial transformation of GPCs, as well as to the subsequent maintenance and progression of malignant phenotype. Since both the TGF-β and wnt/∫β-catenin pathways have been associated with the turnover of normal neural stem cells and GPCs (Feigenson et al., 2009; Mishra et al., 2005), our identification of their tumor progenitor-selective components may provide us an especially compelling set of therapeutic targets.

This comparison revealed a core set of genes dysregulated during both the early and late stages of tumor progression. These transcripts included the selectin ligand CD24, the SIX1 homeodomain transcription factor and its binding partner EYA1, the de-differentiation-associated gene SATB2, and the growth and invasion-associated transcripts MMP3, IGFBP3, CSRP2 and GAP43. By virtue of their marked differential over-expression by TPCs at both early and late stages of glioma progression, these genes appear to comprise a promising set of targets for therapeutic intervention. To define those gene sets preferentially associated with anaplastic progression and invasiveness, we also compared the expression patterns of A2B5+ TPCs derived from HG tumors to those derived from lower grade astrocytomas and oligodendrogliomas. We found that with anaplastic progression, the A2B5+ TPCs upregulated genes associated with the acquisition of mesenchymal phenotype, confirming prior studies using resected glioma tissues (Phillips et al., 2006; Verhaak et al., 2010). Importantly, the enrichment of EMT-associated genes, with a concurrent enrichment of genes involved in motility, all suggest that the A2B5-defined pool includes those cells most associated with both malignant progression and parenchymal invasion.

Our analysis also identified a discrete set of genes that were highly differentially over-expressed during initial and late stages of glioma progression, which thus comprise especially attractive favorable set of therapeutic targets. These include the glycoprotein CD24, a glycophosphatidyl-inositol (GPI)- anchored protein that has been associated with invasiveness and progression in a variety of solid tumors (Baumann et al., 2005). Our data were in accordance with previous reports showing that CD24 protein was overexpressed in both LG and malignant gliomas, and is associated with poor prognosis and increased invasiveness (Deng et al., 2012; Senner et al., 1999).

We also identified the transcription factor SIX1 (Sine oculis) as the second highest differentially expressed transcript by A2B5+ TPCs derived from both LG and HG gliomas. Interestingly, SIX1 overexpression has been described in a number of solid tumors including gliomas, and has been correlated with worse clinical prognosis (Micalizzi et al., 2009), while its mis-expression can result in the transformation of human mammary epithelial cells. In addition, induced Six1 over-expression promotes the proliferation and metastatic dissemination of breast cancer cells, and potentiates TGF-β signaling while inducing an epithelial-mesenchymal transition (EMT) (Micalizzi et al., 2009). Extending these data, we found that SIX1 silencing inhibits the growth, proliferation and survival of A2B5+ TPCs in vitro, and their tumorigenicity in vivo. The likely importance of SIX1 over-expression in A2B5+ glioma TPCs was emphasized by the simultaneous over-expression of EYA1, a principal co-activator of SIX1, at every stage of gliomagenesis, and the subsequent down-regulation of its repressor DACH2, during anaplastic progression (Christensen et al., 2008). Together, these data suggest that the SIX-EYA-DACH transcriptional complex plays a significant, hitherto unrecognized role in glial tumorigenesis.

This study focused on A2B5+ TPCs, since A2B5 expression characterized a discrete population of progenitors that could be identified in normal brain, and which could then be serially tracked throughout disease progression. Yet A2B5-defined TPCs were by no means the sole tumor-initiating cells; A2B5-depleted populations proved tumorigenic as well, particularly those derived from GBM. Rather, the A2B5 pool appeared to identify an especially early-appearing phenotype, which proved virulent; in matched comparisons, the A2B5+ cells propagated new tumors at lower cell doses than did A2B5-depleted cells, and typically exhibited more rapid clonogenic expansion than did A2B5− cells derived from the same tumors. Together, these data suggest that multiple distinct tumor initiating phenotypes, perhaps lineally related but nonetheless antigenically distinct, may coexist in GBM. The A2B5-defined phenotype is notable for appearing in the early stages of gliomagenesis, and persisting at all identified stages of tumor progression. Its selective expression of EMT genes with anaplastic progression suggests its role in tumor virulence, supported by the dominance of the A2B5 phenotype in sustained culture. The dominance of this phenotype in vitro portends both its aggressiveness and chemoresistance in vivo, since A2B5+ glioma cells are preferentially resistant to the nitrosourea CCNU (Balik et al., 2009), one of the few approved chemotherapeutics for glioma.

More broadly, the A2B5-defined phenotype appears to comprise but one of several antigenically distinct glioma-propagating phenotypes, each of which is independently capable of tumor initiation in a naive host. In this regard, it joins CD133 (Singh et al., 2004), CD15/SSEA1 (Son et al., 2009), and integrin-α6 (Lathia et al., 2010), as one of a number of only partially overlapping tumor-initiating phenotypes, that together contribute to much of the virulence of malignant gliomas. Yet the A2B5 phenotype itself may be heterogeneous, so that it remains to be seen whether the A2B5-defined tumor signatures established in our study reflect the dysregulation of specific genes by all A2B5+ TPCs, or whether they instead reflect a specific subpopulation thereof. Given this diversity of tumor-initiating phenotypes within malignant glioma, our challenge is to identify and characterize the smallest possible set of non-overlapping phenotypes that includes all cells capable of independent tumor initiation, and to then separately eliminate each of them, whether serially or concurrently, as a strategy for rational, phenotype-selective and pathway-targeted tumor elimination.

EXPERIMENTAL PROCEDURES

Tissue samples

Tumor samples were graded in accord with WHO guidelines as oligodendroglioma (LG n=4; anaplastic n=4), astrocytoma (diffuse, n=3; anaplastic, n=3), mixed oligoastrocytoma (LG n = 4; anaplastic, n=2), glioblastoma (n=16), ganglioglioma (n=1). Normal epileptic tissue resections obtained from 54 patients were used as controls. Among these, a set of matched gray and white matter-derived A2B5-sorted GPCs were isolated from 4 patients, aged 30 – 46 yrs. All samples were obtained from patients who consented to tissue use, under protocols approved by both the University of Rochester and Johns Hopkins University IRBs.

Cell preparation and isolation

Tissues were dissociated using papain, cultured in cell suspension plates, in serum-free media (SFM) defined as DMEM/F12 media supplemented with N1, 20ng/ml bFGF, EGF and PDGF-AA. Tumor-derived GPCs were then isolated by tissue dissociation followed by A2B5-based cell sorting, both magnetic (MACS) and fluorescent activated (FACS), using previously described protocols (Nunes et al., 2003).

Orthotopic transplantation and analysis

To assess in vivo tumorigenicity, adult immunodeficient mice were injected with TPCs of the sorted phenotypes noted, and killed at various time-points thereafter. All transplantation procedures were reviewed and approved by the Rochester University Committee on Animal Resources. At the time of sacrifice, xenografted brains were cut and donor cells identified by immunolabeling for human nuclear antigen (HNA), human GFAP, or human nestin, as well as Ki67, survivin, and P53. The antibodies used are listed in Table S10.

Differential gene expression and pathway analysis

RNA isolated from A2B5-selected cells was labeled and hybridized to Affymetrix U133+2 arrays. All analyses were performed in R/Bioconductor. Microarray data were preprocessed using robust multichip analysis (RMA) and informative probe sets were subsequently determined using FARMS. Principal component analysis (PCA) and hierarchical clustering were performed on normalized data. Differential gene expression analysis was performed using a linear model approach, employing an empirical Bayesian method for calculation of statistical significance (Bioconductor, limma package). A 3- fold change threshold with significance at 1% FDR (False discovery rate) was generally applied to define differential expression. Pathway analyses were performed using Ingenuity Pathway analysis, as well as several open-source systems.

Clonogenicity, multipotency, cell proliferation and cell death assays

Clonogenicity was assessed 14 days after dissociation of gliomaspheres into single cells and distributed to 96-well plates at 5 – 100 cells/well with 0.2 ml/well of SFM. Multipotency was examined on individual spheres cultured in DMEM/F12/N1 with 1% FBS for up to 12 days. The cells were then fixed and immunostained for GFAP, Olig2, CNP, βIII-tubulin, or MAP-2AB, followed by Alexa-Fluor conjugated secondary antibodies. Cell proliferation and cell cycle analysis were performed using BrdU (30μM) and EdU (10μM) administration followed by respectively, immunocytochemical analysis and Click-it based Edu flow cytometry analysis (Invitrogen) following manufacturer’s instructions. Cell apoptosis and cell death was assessed by flow cytometry analysis of Annexin V (BD) and DAPI respectively.

Generation, validation and assessment of Six1 knock-down lentivirus

A set of five lentiviral shRNAi vectors with distinct target sequences was purchased from Open Biosystems. Six1-induced silencing constructs were validated by transfection of multiple glioma cell lines, with subsequent q-PCR and Western immunoblot detection of Six1 mRNA and protein expression. Viral production, cell transduction, validation of knock-down, and assessment of effects thereof on glioma cell proliferation and death are all described in the Supplementary Methods.

Statistical analysis

Statistical analysis for all experiments including more than two groups were performed using a one-way or two-way ANOVA followed by post hoc comparisons using Tukey multiple comparison test. For experiments with two groups, we used Student’s t test.

Supplementary Material

Supplemental

HIGHLIGHTS.

  • Gene expression by normal versus neoplastic A2B5+ glioma tumor progenitor cells (TPC)

  • A2B5 TPC transit from proneural to mesenchymal phenotype with glioma progression

  • SIX1 and EYA1 are overexpressed by A2B5+ TPCs at all stages of glioma progression

  • Inhibition of SIX1 suppresses glioma TPC expansion in vitro and in vivo

ACKNOWLEDGMENTS

This work was supported by grants from the Adelson Medical Research Foundation, The New York State Stem Cell Research Board (NYSTEM), the James S. McDonnell Science Foundation, Sanofi- Aventis, NINDS R01NS39559, and K08NS055851. We thank Drs. Webster Pilcher, Howard Silberstein and Edward Vates for arranging tissue donation, and Tricia Protack, Adam Cornwell, Matthew J. Kaule, Alex Zielke, Li Xiaojie, Marissa Dombovy-Johnson, and Andrew Bennett for technical assistance.

Footnotes

ACCESSION NUMBERS

Microarray data described herein are available at the NCBI Gene Expression Omnibus under the following link: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=bbehvogwcaoiidc&acc=GSE29796

SUPPLEMENTAL INFORMATION

Supplemental information includes 10 figures, 10 tables, Supplemental Experimental Procedures and Supplemental References.

REFERENCES

  1. Alcantara Llaguno S, Chen J, Kwon CH, Jackson EL, Li Y, Burns DK, Alvarez-Buylla A, Parada LF. Malignant astrocytomas originate from neural stem/progenitor cells in a somatic tumor suppressor mouse model. Cancer Cell. 2009;15:45–56. doi: 10.1016/j.ccr.2008.12.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Assanah M, Lochhead R, Ogden A, Bruce J, Goldman J, Canoll P. Glial progenitors in adult white matter are driven to form malignant gliomas by platelet-derived growth factor-expressing retroviruses. J Neurosci. 2006;26:6781–6790. doi: 10.1523/JNEUROSCI.0514-06.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Balik V, Mirossay P, Bohus P, Sulla I, Mirossay L, Sarissky M. Flow cytometry analysis of neural differentiation markers expression in human glioblastomas may predict their response to chemotherapy. Cellular and molecular neurobiology. 2009;29:845–858. doi: 10.1007/s10571-009-9366-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bao S, Wu Q, McLendon RE, Hao Y, Shi Q, Hjelmeland AB, Dewhirst MW, Bigner DD, Rich JN. Glioma stem cells promote radioresistance by preferential activation of the DNA damage response. Nature. 2006;444:756–760. doi: 10.1038/nature05236. [DOI] [PubMed] [Google Scholar]
  5. Baumann P, Cremers N, Kroese F, Orend G, Chiquet-Ehrismann R, Uede T, Yagita H, Sleeman JP. CD24 expression causes the acquisition of multiple cellular properties associated with tumor growth and metastasis. Cancer Res. 2005;65:10783–10793. doi: 10.1158/0008-5472.CAN-05-0619. [DOI] [PubMed] [Google Scholar]
  6. Christensen KL, Patrick AN, McCoy EL, Ford HL. The six family of homeobox genes in development and cancer. Adv Cancer Res. 2008;101:93–126. doi: 10.1016/S0065-230X(08)00405-3. [DOI] [PubMed] [Google Scholar]
  7. Deng J, Gao G, Wang L, Wang T, Yu J, Zhao Z. CD24 expression as a marker for predicting clinical outcome in human gliomas. J Biomed Biotechnol. 2012;2012:517172. doi: 10.1155/2012/517172. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Di Benedetto M, Bieche I, Deshayes F, Vacher S, Nouet S, Collura V, Seitz I, Louis S, Pineau P, Amsellem-Ouazana D, et al. Structural organization and expression of human MTUS1, a candidate 8p22 tumor suppressor gene encoding a family of angiotensin II AT2 receptor-interacting proteins, ATIP. Gene. 2006;380:127–136. doi: 10.1016/j.gene.2006.05.021. [DOI] [PubMed] [Google Scholar]
  9. Earl J, Yan L, Vitone LJ, Risk J, Kemp SJ, McFaul C, Neoptolemos JP, Greenhalf W, Kress R, Sina-Frey M, et al. Evaluation of the 4q32-34 locus in European familial pancreatic cancer. Cancer Epidemiol Biomarkers Prev. 2006;15:1948–1955. doi: 10.1158/1055-9965.EPI-06-0376. [DOI] [PubMed] [Google Scholar]
  10. Feigenson K, Reid M, See J, Crenshaw EB, 3rd, Grinspan JB. Wnt signaling is sufficient to perturb oligodendrocyte maturation. Mol Cell Neurosci. 2009;42:255–265. doi: 10.1016/j.mcn.2009.07.010. [DOI] [PubMed] [Google Scholar]
  11. Freije WA, Castro-Vargas FE, Fang Z, Horvath S, Cloughesy T, Liau LM, Mischel PS, Nelson SF. Gene expression profiling of gliomas strongly predicts survival. Cancer Res. 2004;64:6503–6510. doi: 10.1158/0008-5472.CAN-04-0452. [DOI] [PubMed] [Google Scholar]
  12. Galli R, Binda E, Orfanelli U, Cipelletti B, Gritti A, De Vitis S, Fiocco R, Foroni C, Dimeco F, Vescovi A. Isolation and characterization of tumorigenic, stem-like neural precursors from human glioblastoma. Cancer Res. 2004;64:7011–7021. doi: 10.1158/0008-5472.CAN-04-1364. [DOI] [PubMed] [Google Scholar]
  13. Hedge TA, Mason I. Expression of Shisa2, a modulator of both Wnt and Fgf signaling, in the chick embryo. Int J Dev Biol. 2008;52:81–85. doi: 10.1387/ijdb.072355th. [DOI] [PubMed] [Google Scholar]
  14. Hemmati HD, Nakano I, Lazareff JA, Masterman-Smith M, Geschwind DH, Bronner-Fraser M, Kornblum HI. Cancerous stem cells can arise from pediatric brain tumors. Proc Natl Acad Sci U S A. 2003;100:15178–15183. doi: 10.1073/pnas.2036535100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Ignatova TN, Kukekov VG, Laywell ED, Suslov ON, Vrionis FD, Steindler DA. Human cortical glial tumors contain neural stem-like cells expressing astroglial and neuronal markers in vitro. Glia. 2002;39:193–206. doi: 10.1002/glia.10094. [DOI] [PubMed] [Google Scholar]
  16. Ikushima H, Todo T, Ino Y, Takahashi M, Miyazawa K, Miyazono K. Autocrine TGF-beta signaling maintains tumorigenicity of glioma-initiating cells through Sry-related HMG-box factors. Cell Stem Cell. 2009;5:504–514. doi: 10.1016/j.stem.2009.08.018. [DOI] [PubMed] [Google Scholar]
  17. Jackson EL, Garcia-Verdugo JM, Gil-Perotin S, Roy M, Quinones-Hinojosa A, VandenBerg S, Alvarez-Buylla A. PDGFR alpha-positive B cells are neural stem cells in the adult SVZ that form glioma-like growths in response to increased PDGF signaling. Neuron. 2006;51:187–199. doi: 10.1016/j.neuron.2006.06.012. [DOI] [PubMed] [Google Scholar]
  18. Langford LA, Piatyszek MA, Xu R, Schold SC, Jr, Shay JW. Telomerase activity in human brain tumours. Lancet. 1995;346:1267–1268. doi: 10.1016/s0140-6736(95)91865-5. [DOI] [PubMed] [Google Scholar]
  19. Lathia JD, Gallagher J, Heddleston JM, Wang J, Eyler CE, Macswords J, Wu Q, Vasanji A, McLendon RE, Hjelmeland AB, Rich JN. Integrin alpha 6 regulates glioblastoma stem cells. Cell Stem Cell. 2010;6:421–432. doi: 10.1016/j.stem.2010.02.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. 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]
  21. Lindberg N, Kastemar M, Olofsson T, Smits A, Uhrbom L. Oligodendrocyte progenitor cells can act as cell of origin for experimental glioma. Oncogene. 2009;28:2266–2275. doi: 10.1038/onc.2009.76. [DOI] [PubMed] [Google Scholar]
  22. Liu C, Sage JC, Miller MR, Verhaak RG, Hippenmeyer S, Vogel H, Foreman O, Bronson RT, Nishiyama A, Luo L, Zong H. 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]
  23. Micalizzi DS, Christensen KL, Jedlicka P, Coletta RD, Baron AE, Harrell JC, Horwitz KB, Billheimer D, Heichman KA, Welm AL, et al. The Six1 homeoprotein induces human mammary carcinoma cells to undergo epithelial-mesenchymal transition and metastasis in mice through increasing TGF-beta signaling. J Clin Invest. 2009;119:2678–2690. doi: 10.1172/JCI37815. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Midorikawa Y, Tsutsumi S, Taniguchi H, Ishii M, Kobune Y, Kodama T, Makuuchi M, Aburatani H. Identification of genes associated with dedifferentiation of hepatocellular carcinoma with expression profiling analysis. Jpn J Cancer Res. 2002;93:636–643. doi: 10.1111/j.1349-7006.2002.tb01301.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Mishra L, Derynck R, Mishra B. TGF-beta signaling in stem cells and cancer. Science. 2005;310:68–71. doi: 10.1126/science.1118389. [DOI] [PubMed] [Google Scholar]
  26. Nishide K, Nakatani Y, Kiyonari H, Kondo T. Glioblastoma formation from cell population depleted of Prominin1-expressing cells. PLoS One. 2009;4:e6869. doi: 10.1371/journal.pone.0006869. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Nishiyama A, Komitova M, Suzuki R, Zhu X. Polydendrocytes (NG2 cells): multifunctional cells with lineage plasticity. Nat Rev Neurosci. 2009;10:9–22. doi: 10.1038/nrn2495. [DOI] [PubMed] [Google Scholar]
  28. Nunes MC, Roy NS, Keyoung HM, Goodman RR, McKhann G, 2nd, Jiang L, Kang J, Nedergaard M, Goldman SA. Identification and isolation of multipotential neural progenitor cells from the subcortical white matter of the adult human brain. Nat Med. 2003;9:439–447. doi: 10.1038/nm837. [DOI] [PubMed] [Google Scholar]
  29. Ogden AT, Waziri AE, Lochhead RA, Fusco D, Lopez K, Ellis JA, Kang J, Assanah M, McKhann GM, Sisti MB, et al. Identification of A2B5+ CD133− tumor-initiating cells in adult human gliomas. Neurosurgery. 2008;62:505–514. doi: 10.1227/01.neu.0000316019.28421.95. discussion 514-505. [DOI] [PubMed] [Google Scholar]
  30. Pandey RN, Rani R, Yeo EJ, Spencer M, Hu S, Lang RA, Hegde RS. The Eyes Absent phosphatase-transactivator proteins promote proliferation, transformation, migration, and invasion of tumor cells. Oncogene. 2010;29:3715–3722. doi: 10.1038/onc.2010.122. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Patani N, Jiang W, Mansel R, Newbold R, Mokbel K. The mRNA expression of SATB1 and SATB2 in human breast cancer. Cancer Cell Int. 2009;9:18. doi: 10.1186/1475-2867-9-18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Patel YC. Somatostatin and its receptor family. Front Neuroendocrinol. 1999;20:157–198. doi: 10.1006/frne.1999.0183. [DOI] [PubMed] [Google Scholar]
  33. 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]
  34. Phillips HS, Kharbanda S, Chen R, Forrest WF, Soriano RH, Wu TD, Misra A, Nigro JM, Colman H, Soroceanu L, et al. Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis. Cancer Cell. 2006;9:157–173. doi: 10.1016/j.ccr.2006.02.019. [DOI] [PubMed] [Google Scholar]
  35. Pincus DW, Keyoung HM, Harrison-Restelli C, Goodman RR, Fraser RA, Edgar M, Sakakibara S, Okano H, Nedergaard M, Goldman SA. FGF-2/BDNF-associated maturation of new neurons generated from adult human subependymal cells. Ann Neurol. 1998;43:576–585. doi: 10.1002/ana.410430505. [DOI] [PubMed] [Google Scholar]
  36. Pouly S, Becher B, Blain M, Antel JP. Expression of a homologue of rat NG2 on human microglia. Glia. 1999;27:259–268. doi: 10.1002/(sici)1098-1136(199909)27:3<259::aid-glia7>3.0.co;2-8. [DOI] [PubMed] [Google Scholar]
  37. Pulvirenti T, Van Der Heijden M, Droms LA, Huse JT, Tabar V, Hall A. Dishevelled 2 signaling promotes self-renewal and tumorigenicity in human gliomas. Cancer Res. 2011;71:7280–7290. doi: 10.1158/0008-5472.CAN-11-1531. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Sanai N, Tramontin AD, Quinones-Hinojosa A, Barbaro NM, Gupta N, Kunwar S, Lawton MT, McDermott MW, Parsa AT, Manuel-Garcia Verdugo J, et al. Unique astrocyte ribbon in adult human brain contains neural stem cells but lacks chain migration. Nature. 2004;427:740–744. doi: 10.1038/nature02301. [DOI] [PubMed] [Google Scholar]
  39. Senner V, Sturm A, Baur I, Schrell UH, Distel L, Paulus W. CD24 promotes invasion of glioma cells in vivo. J Neuropathol Exp Neurol. 1999;58:795–802. doi: 10.1097/00005072-199908000-00002. [DOI] [PubMed] [Google Scholar]
  40. Shih AH, Holland EC. PDGF and glial tumorigenesis. Cancer Lett. 2006;232:139–147. doi: 10.1016/j.canlet.2005.02.002. [DOI] [PubMed] [Google Scholar]
  41. Sim FJ, Lang JK, Waldau B, Roy NS, Schwartz TE, Pilcher WH, Chandross KJ, Natesan S, Merrill JE, Goldman SA. Complementary patterns of gene expression by human oligodendrocyte progenitors and their environment predict determinants of progenitor maintenance and differentiation. Ann Neurol. 2006;59:763–779. doi: 10.1002/ana.20812. [DOI] [PubMed] [Google Scholar]
  42. Sim FJ, McClain CR, Schanz SJ, Protack TL, Windrem MS, Goldman SA. CD140a identifies a population of highly myelinogenic, migration-competent and efficiently engrafting human oligodendrocyte progenitor cells. Nat Biotechnol. 2011;29:934–941. doi: 10.1038/nbt.1972. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Singh SK, Hawkins C, Clarke ID, Squire JA, Bayani J, Hide T, Henkelman RM, Cusimano MD, Dirks PB. Identification of human brain tumour initiating cells. Nature. 2004;432:396–401. doi: 10.1038/nature03128. [DOI] [PubMed] [Google Scholar]
  44. Son MJ, Woolard K, Nam DH, Lee J, Fine HA. SSEA-1 is an enrichment marker for tumor-initiating cells in human glioblastoma. Cell Stem Cell. 2009;4:440–452. doi: 10.1016/j.stem.2009.03.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Stupp R, Mason WP, van den Bent MJ, Weller M, Fisher B, Taphoorn MJ, Belanger K, Brandes AA, Marosi C, Bogdahn U, et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med. 2005;352:987–996. doi: 10.1056/NEJMoa043330. [DOI] [PubMed] [Google Scholar]
  46. Tchoghandjian A, Baeza N, Colin C, Cayre M, Metellus P, Beclin C, Ouafik L, Figarella-Branger D. A2B5 cells from human glioblastoma have cancer stem cell properties. Brain Pathol. 2009 doi: 10.1111/j.1750-3639.2009.00269.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. 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]
  48. 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]

Associated Data

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

Supplementary Materials

Supplemental
NIHMS740097-supplement-Supplemental.pdf (1MB, pdf)

RESOURCES