Characteristic Features of Stem Cells in Glioblastomas: From Cellular Biology to Genetics

Abstract

Glioblastoma is the most common type of primary brain tumor in adults and is among the most lethal and least successfully treated solid tumors. Recently, research into the area of stem cells in brain tumors has gained momentum. However, due to the relatively new and novel hypothesis that a subpopulation of cancer cells in each malignancy has the potential for tumor initiation and repopulation, the data in this area of research are still in its infancy. This review article is aimed at attempting to bring together research carried out so far in order to build an understanding of glioblastoma stem cells (GSCs). Initially, we consider GSCs at a morphological and cellular level, and then discuss important cell markers, signaling pathways and genetics. Furthermore, we highlight the difficulties associated with what some of the evidence indicates and what collectively the studies contribute to further defining the interpretation of GSCs.

INTRODUCTION

Recently, research into the area of stem cells in brain tumors has gained momentum. Indeed, the first evidence for the existence of any subpopulation of cancer stem cells (CSCs) within a tumor came in 1994 from studying acute myeloid leukemia (92). As such, due to the relatively new and novel experimental evidence that CSCs exist and have the potential for tumor initiation and repopulation, the data in this area of research are still in its infancy. Thus, it is the purpose of this review to attempt to bring together some of the research in order to attempt to define the characteristic features of stem cells in glioblastomas, covering the morphological features, cellular biology and genetics.

There are two main reasons why a review is useful. First, interest in this area of oncology has increased dramatically in recent years. This is evidential from the amount of research being produced and published on the subject of glioblastoma stem cells (GSCs). Using a simple keyword search of “glioblastoma stem cell” in key journal search engines produced thousands of related articles in 2011: ScienceDirect, 5541; MEDLINE, 7266; and PubMed, 832. There was a significant increase compared with the same search dated up to and including 1992: ScienceDirect, 496; MEDLINE, 832; PubMed, 44.

Second, glioblastoma is the most common type of primary brain tumor in adults and is among the most lethal and least successfully treated solid tumors (46), occurring in 2–3 cases per 100 000 people in Europe and North America (167). The median survival time of patients with glioblastoma multiforme (GBM) following treatment is 12 months, with only 3–5% of patients surviving longer than 5 years 32, 56. Compared with the advances in the treatment of other types of tumors, the poor prognosis for GBM patients has improved minimally over decades, underscoring the challenges and difficulties in effectively detecting and treating these fatal cancers (27).

Such a poor prognosis has several causes, including the extensive dissemination of tumor cells, making total surgical resection almost, if not completely, impossible; the tumor is highly refractory to available chemotherapeutic and radiation treatments and there is still insufficient understanding of glioblastomas. This review article is therefore aimed at attempting to bring together research carried out so far in order to build an understanding of GSCs.

The nomenclature of stem cells in glioblastomas remains unresolved and thus far there are no standardized criteria for identifying them. There are a number of terms used to denote the same concept including GSC, glioma tumor initiating cells and brain tumor stem‐like cells. The terminology reflects some shared characteristics with normal stem cells, especially adult somatic stem cells, including the capacities for self‐renewal, differentiation and maintained proliferation (153). Furthermore, it is known that cancer stem cells are distinct from the cell of origin, indeed the phenotype of the cell of origin may differ substantially from that of the cancer stem cell (165). It is suggested that cancer stem cells do not necessarily originate from the transformation of normal stem cells, but rather from restricted progenitors or more differentiated cells that have acquired self‐renewing capacity (166). In this article, the term GSC is used.

MORPHOLOGICAL AND CELLULAR BIOLOGY OF GSCs, AND DIFFICULTY IN LOCATING GSC NICHES

Table 1 shows that there are a number of morphological and biological features that can distinguish GSCs. Despite the strong evidence for the existence of the three main areas of neuronal production, more recently there have been several reports of adult neurogenesis in other regions of the brain, including the striatum 33, 102, the amygdala 13, 41, the hypothalamus 74, 176 and the substantia nigra (183), as well as in the brain stem (11) and olfactory tubercle (12). However, there are also studies which have failed to confirm these findings 26, 30, 44 or only confirmed the finding following damage or pharmacological intervention (57). If the origin of neural stem cells can be located, then a more precise strike could be dealt when delivering treatment for glioblastomas. However, there are two main problems which would still have to be resolved, even if neural stem cell niches are located.

Table 1.

Morphological and biological characteristics of GSCs. Abbreviations: GBM = glioblastoma multiforme; GSC = glioblastoma stem cell

Feature	Evidence of existence in GSCs	Reference
Origin of glioblastomas	Cerebral hemispheres, typically at the cortical/subcortical interface and spread through the white matter, occasionally to the opposite hemisphere through the corpus callosum. GBM cells are extremely infiltrative, often migrating along the basement membrane of blood vessels or along myelinated white matter.	14, 101
Origin of neural stem cells	Three sites of neuronal production in the adult mammalian brain; the subgranular zone of the dentate gyrus of the hippocampus, the subventricular zone which lines the lateral walls of the ventricles and the subependymal zone of the spinal cord	2, 47, 48, 81
Migrating ability	Possess the ability to migrate and differentiate after transplantation into a rodent brain. It has also been shown that neuronal precursors can migrate to a damaged environment within the brain which has resulted from ischemia or the presence of a tumor.	1, 67
Cellular biology	GSCs have many of the same characteristics of normal stem cells such as the formation of floating neurospheres, more G0/G1 phase cells, more resistance to hypoxia, irradiations and some chemotherapeutics, higher colony‐formation ability, and high capacity to develop into multiple lineages.	(124)
Histology	Indistinguishable from normal stem cells	(165)
Karyotype analysis	GSCs have been shown to have aneuploidy of specific autosomes and sex chromosomes; specifically monosomy 1, trisomy 7 and 22, and disomy X.	(49)
Proteins	Express high levels of ATP‐binding cassette transporters and proteins related to drug resistance, such as P‐glycoprotein, multidrug resistant protein (MRP)‐1, MRP3, MRP5, breast cancer drug‐resistance protein, and glutathione‐S‐transferase	124, 180
Enzymes	Expression of matrix metalloproteinase‐13 was specifically expressed in GSCs, indicating they have highly invasive potential.	(76)
Number of GSCs needed for tumor formation	Analysis of mouse brains following CD133+ engraftment revealed that as few as 100 CD133+ cells were sufficient for the formation of human brain tumors in mice.	(143)

First, neural stem cells differ according to the region of brain in which they are located. The detection of a genetic signature specific to cerebella vs. neocortical neural stem cells suggests that morphologically indistinguishable stem cells from different brain regions are unique (138). Two observable differences in neural progenitor cells (NPCs) which relate to their origin are: (i) NPCs from the rostral part of the brain tended to proliferate faster than those from the caudal part; and (ii) the NPCs from the diencephalon and mesencephalon gave rise to more tyrosine hydroxylase‐positive neurons than did those from the telencephalon and rhombencephalon (70). If neural stem cells differ according to the region of the brain in which they are located, then it may also be the case that GSCs also differ according to the region of brain in which they are located. Thus it may be that, depending on where the glioblastoma precisely originates, certain types of GSCs may have specific mutations which are unique to their location of origin.

Second, evidence suggests that neural stem cells can migrate, a feature which may also be attributed to GSCs. The extent of this is considerable as it has also been shown that mesenchymal stem cells (which can transdifferentiate into neural cells in vitro under the influence of matrix molecules and growth factors present in neurogenic niches) have the potential to migrate long distances, such as from the subventricular zone (SVZ) toward the olfactory bulb through the rostral migratory stream (34). However, evidence shows that the migration mode of SVZ progenitor cells can be controlled by environmental cues and it was further proposed that subcortical white matter tracts present structural and/or biochemical properties favorable to progenitor cell migration (25). Manipulation of the local environment of stem cell niches and therefore GSC niches could provide a gateway into controlling both the initial damage and future spread of glioblastomas.

However, despite the difficulties associated with pinpointing GSC niches, recently, there have been a collection of reports which indicate two particular areas where GSCs may be found. Bao et al was the first to show that GSCs develop a special relationship with the surrounding vasculature, as it was found that CD 133+ human glioblastoma cells consistently secrete high levels of vascular endothelial growth factor (VEGF) and that this might contribute to their tumor‐initiating capacity (6). Soon after, Calabrese et al demonstrated that GSC fractions were located next to capillaries in brain tumors and that endothelial cells interact selectively with these GSCs in culture and supply secreted factors that maintain these cells in self‐renewing and undifferentiated states (22). They also found that targeting a brain tumor with bevacizumab (a drug that blocks angiogenesis) reduces the number of cancer stem cells in treated tumors (22).

Furthermore, Heddleston et al, in examining the role of hypoxia in regulating tumor cell plasticity, found that a hypoxic environment promotes the self‐renewal capability of the stem and nonstem population (65). It was also found to promote a more stem‐like phenotype in the nonstem population, with increased neurosphere formation as well as up‐regulation of important stem cell factors, such as OCT4, NANOG and c‐MYC (65). A more recent study has also shown that in exposing GSCs to a hypoxic state (1% oxygen) for 72 h resulted in increased expression of multiple genes including lysyl oxidase (LOX), VEGF, hypoxia inducible gene 2 (HIG2) and Krüpple‐like factor 4 (Klf4) (9). Klf4 is particularly important as it plays an important role in maintaining embryonic stem cells and preventing their differentiation. Klf4 also contributes to the down‐regulation of p53 transcription. The study by Bar et al also goes on to show that by using digoxin to target the hypoxic response in GBM, via the inhibition of hypoxia inducible factor (HIF), growth of glioma cell lines in vitro and in vivo can be slowed, suggesting that GSC responses to hypoxia can be suppressed (9). Seidel et al also showed that hypoxia plays a key role in the regulation of the tumor stem cell phenotype through HIG2 alpha, indicating that tumor cells are also maintained within a hypoxic niche and could thus provide a GSC niche (136).

CELL MARKERS OF GSCs

Much of the research regarding cell markers has been directed toward CD133, as seen in Table 2. Early research showed that within glioblastomas, a subpopulation of cells existed. This subpopulation which represented a minority of the tumor was identified by expression of the cell surface marker CD133. It was confirmed as a brain tumor stem cell in 2003 and it was the first marker of GSCs (142). Following its discovery, CD133 (a pentaspan transmembrane glycoprotein cell surface molecule) has been widely used as a marker for enrichment of a GSC population (26). As such, CD133 continues to be used as the primary cell marker of GSCs. In reviewing Table 2, it is of note that there is still limited data on cell markers of GSCs, thus it is difficult to conclude that any one marker will provide greater benefit in aiding the location of GSCs. Furthermore, as Table 2 shows, in ascertaining the percentage of GSCs that express marker per tumor, all samples used have been cultured in some way which further adds difficulty when making comparisons between cell markers.

Table 2.

Cell markers of GSCs, percentage expression per sample size and percentage of GSCs that express marker per tumor. Abbreviations: GSC = glioblastoma stem cell; N/A = data not available

Cell marker	Sample size	Percentage of expression per sample size	Percentage of GSCs that express marker per tumor	Methodology of preparing specimens	Study
Cell surface markers
CD133	25	N/A	11.2–18.4	Analyzed using flow cytometry	(113)
	4	N/A	19–29	In vitro primary sphere formation assays performed on all uncultured cells prior to flow cytometric quantification	(143)
	24	54	1.7–47.1	Enzymatically dissociated into single cells and cultured; sorted by flow cytometric analysis	(146)
	7	N/A	2.0–94.4	Biopsies were dissociated into single cells and cultured; sorted by flow cytometric analysis.	(164)
	16	75	N/A	Immunohistochemical staining using mouse antihuman CD133/1‐biotin MAb, CD133/2‐biotin MAb, and CD133/1MAb antibody. Signals were counted in eight random fields of each single tumor specimen.	(159)
	125	78.4	N/A	Immunohistochemical staining using anti‐Nestin monoclonal antibody. The number of positive‐staining cells showing immunoreactivity on the cell membranes and cytoplasm in 10 representative microscopic fields was counted and the percentage of positive cells was calculated.	(182)
	47	96	N/A	Immunohistochemical staining using mouse monoclonal anti‐CD133 antibody	(181)
Integrin α6	7	100	1.2–41.1	Biopsy was used immediately after dissociation or after transient xenograft passage in immunocompromised mice. Cells were sorted by fluorescence‐activated cell sorting or magnetic bead separation	(94)
SSEA/CD15	24	96	1.4–84.2	Enzymatically dissociated into single cells and cultured; sorted by flow cytometric analysis	(146)
	7	N/A	3.0–15.2	Biopsies were dissociated into single cells and cultured; sorted by flow cytometric analysis	(164)
L1CAM	3	100	3.8–6.3	Matched glioma cell populations were isolated and cultured and then subjected to fluorescence‐activated cell sorting	(7)
CXCR4	7	100	1.1–40.2	Biopsies were dissociated into single cells and cultured, followed by flow cytometric analysis	(164)
A2B5	25	N/A	57.9–65.5	Analyzed using flow cytometry	(113)
Intracellular markers
Nestin	125	82.4	N/A	Immunohistochemical staining using anti‐Nestin monoclonal antibody. The number of positive‐staining cells showing immunoreactivity on the cell membranes and cytoplasm in 10 representative microscopic fields was counted and the percentage of positive cells was calculated.	(182)
	7	100	10–97.5	Samples were cultured in Dulbecco's modified Eagle's medium; immunocytochemistry was used for quantification of the cells positive for a specific marker	(52)
Sox‐2	7	100	31–85.5	Samples were cultured in Dulbecco's modified Eagle's medium; immunocytochemistry was used for quantification of the cells positive for a specific marker	(52)
	11	90	N/A	Immunohistochemistry was performed; the proportion of Sox2‐positive cells was determined by counting all cell nuclei as well as nuclei stained for Sox2 in three randomly selected high‐power fields in the tumor core of each sample	(134)

A problem, however, in using CD133 as the gold standard cell marker for glioblastomas, is that many normal cells express CD133 as well. This fact potentially limits the use of CD133 as a target, as the ability to discriminate GSCs from non‐GSCs is not absolute (27). This was confirmed by Tchoghandjian et al who isolated stem cells from glioblastomas expressing the surface marker A2B5 (157). This subpopulation was then divided into two groups using the CD133 antigen into A2B5+/CD133+ and A2B5+/CD133− stem cells, both of which were able to generate tumors when grafted in to nude mice. A2B5 is a cell surface ganglioside that marks neural precursor cells in the adult human brain (112). As Table 2 shows, A2B5 is also a useful GSC marker, especially it would seem in the absence of CD133.

Furthermore, the point of course in attaining a reliable GSC marker is so that it can be used as a target in clinical therapy. Worryingly, it has been found in at least one study that when GBM patients are classified into two groups (CD133‐high group where the CD133+ cell ratio is <3% and CD133‐low group where the CD133− cell ratio is >3%), the CD133‐low GBMs showed more aggressive morphologies as determined by magnetic resonance imaging and unique gene expression patterns that were related to a worse prognosis (82). Kim et al also found no relationship between the level of stem cell marker expression of CD133, SSEA or nestin and prognostic implications, and in fact found that patients with GBM who were CD133+ had a slightly longer survival rate than did the CD133− group (89). A study by Wang et al showed that using CD133− cells derived from six different patients was tumorgenic when implanted into rat brains and from these tumors CD133 positive cells could be obtained (169). Therefore, it is likely that a more primitive cancer stem cell phenotype precedes the CD133+ GSCs.

As with the cell marker CD133, as Table 2 indicates, SSEA‐1 does not enrich for a GSC population in every tumor either. The antigen SSEA‐1 is a fructose‐containing trisaccharide, which is highly expressed on embryonic stem cells in the developing brain and adult SVZ (24). SSEA‐1+ cells have increased expression of stem cell genes, such as Sox‐2 and Bmi1, and are capable of self‐renewal and multilineage differentiation (146). Furthermore, preliminary data have shown that in some tumors, both SSEA‐1−/CD133+ cells and SSEA‐1+/CD133− cells are much more tumorigenic as compared with SSEA‐1−/CD133− cells (146). Therefore, like A2B5, SSEA‐1 should prove to be a useful enrichment marker for GSCs especially for those that are CD133 negative.

One of the more recent cell markers to be analyzed is CXCR4 and the limited data collected so far on this particular cell marker are shown in Table 2. CXCR4 is a G‐protein coupled chemokine receptor which is known to be widely expressed on different cell types (43). Within glioblastomas, CXCR4 expression has been found to increase with increasing grade and colocalizes to regions within these tumors where its interaction may contribute to angiogenesis (128). Furthermore, cancer cells with high levels of CXCR4 are more likely to metastasize (4). It has only been recently reported that CXCR4 is overexpressed in primary glioblastoma progenitor cells and the stimulation of CXCR4 specific ligand, CXCL12, induces a significant proliferative response in the cancer stem cell but not in corresponding differentiated cells (38). Therefore, the role of CXCR4 as a cell marker for GSCs may be more significant than Table 2 suggests.

In the same light as CXCR4, there has been limited investigation on the cell marker L1CAM. This particular marker is a cell adhesion molecule derived from the immunoglobulin family (126). The protein cell marker has been found to interact with several extracellular matrix proteins as well as epidermal growth factor receptor, neuropilin‐1, and α5β1 and αvβ3 integrins 104, 133. Interestingly, Cheng et al (28) found that although a subpopulation of cancer cells inside tumors expressed L1CAM, the invasive fronts of primary GBM showed many more cells with high levels of L1CAM. Immunofluorescent staining further confirmed that many more L1CAM‐positive cells were localized in the invasive fronts than the center region of GBM (28).

Also to date, limited data have been collected on the integrin α6, as shown in Table 2. Integrins are heterodimeric proteins with alpha and beta subunits (35) which make up a large family of receptors that mediate the adhesive interactions of cells (107). Similarly to the L1CAM protein, integrin α6 can mostly be found on the invasive fronts of GBM. Specifically, it was found that 60% of integrin α6‐positive cells were located within 5 µm of a blood vessel as compared with only 10% of total tumor cells (94).

Finally, as indicated by Table 2, intracellular markers Sox‐2 and nestin show promise as reliable GSC markers, but their use is limited because they are intracellular, making the study of live cells difficult. To use either of these markers while cells are still alive and viable would require the use of the reporter system. Furthermore, the most recent evidence suggests that Sox‐2 may not be a reliable GSC marker because it can be found in high frequencies even in low‐grade gliomas and was not found to be linked to patient survival either (168). Evidence by Wan et al throws into doubt the usefulness of the findings that Sox‐2 expression can be found in many glioblastoma tumors as Table 2 highlights. Sox‐2 is a member of the Sry‐related High Mobility Group box family of transcription factors (61) and whose overexpression can induce pluripotency in both mice and human somatic cells 154, 155. More recently, Gangemi et al showed that silencing of the Sox2 in freshly derived GSC stopped proliferation and the resulting cells' tumorigenicity in immunodeficient mice (52). Interestingly, a study by Sun and Zhang showed that arsenic trioxide can induce the apoptosis of GSCs, at least partly through down‐regulation of Sox‐2 (152). Therefore, Sox‐2 usefulness as a GSC marker may no longer be as credible as once perceived, but despite this, its importance in glioblastoma formation is yet to be fully understood and therefore is not to be underestimated.

Nestin is also an intracellular marker, but shows more promise as a reliable marker of GSCs. Nestin is an intermediate filament protein involved in the organization of the cytoskeleton, but it has also been implicated in cell signaling, organogenesis and cell metabolism (45). Nestin was one of the first cell markers to be detected originally by the monoclonal antibody Rat.401 and by immunochemistry and double‐label fluorescence‐activated cell sorting (FACS), where it was shown to be expressed in rat CNS stem cells 42, 68, 96.

Table 2 incorporates cell markers which have been measured according to whether or not any of the GSCs within a number of tumors expresses the particular marker and also how many of those GSCs within a single tumor express the marker. Those cell markers included, however, are by no means exhaustive and other markers of GSCs (which have not been measured in the same way as those present in Table 2) have been found as potential markers for GSCs. For example, Kondo et al used the fluorescent dye Hoechst 33342 and flow cytometry to distinguish a side population (a subpopulation of cells that is distinct in biological characteristics compared with the main population) from glioma cell lines. Kondo et al note that this method of isolating the GSCs provides a simple and general strategy for identifying the cells (91). Similarly, Harris et al reported that GSCs are enriched in the side population cells and that these cells show enhanced self‐renewal and multipotentiality compared with nonside population cells from the same tumor (64). The side population was sorted from the nonside population by staining tumorsphere cells with Hoechst 33342 dye and sorting via FACS (64). However, more recently Broadley et al show that sphere formation in GBM cell lines and primary GBM cells enriches for a CSC‐like phenotype of increased self‐renewal gene expression in vitro and increased tumor initiation in vivo (16). However, the side population was absent from all sphere cultures. Direct isolation of the side population from the GBM lines did not enrich for stem‐like activity in vitro, and tumor‐initiating activity was lower in sorted side population compared with non‐\side population and parental cells (16).

Other GSC markers that have been used include musashi, an RNA binding protein which has been shown to be expressed in glioblastomas and is known in mammals to control neural stem cell homeostasis, differentiation and tumorigenesis, but expression was significantly lower than that of nestin when compared in the same study (149). In a study by Kang and Kang, GBM cancer cells were exposed to 1,3‐bis(2chloroethyl)‐1‐nitrosourea (BCNU) and it was found that a small population of cells survived and proliferated. Some of these subpopulations contained stem‐like cells, which expressed CD133, CD117, CD90, CD71 and CD45 cell‐surface markers and had the capacity for multipotency and when transplanted into mice induced tumors (86). Furthering their studies, Kang et al found that radio‐resistant subpopulations of GBM stem cells also expressed the cell surface marker CD117 as well as CD133, CD71 and CD45 (87).

SIGNALING PATHWAYS COMMONLY FOUND IN GSCs

Key signaling pathways as potential GSC targets are illustrated in Table 3 and indicate the receptors which can activate them and the genes they affect. Understanding the signaling pathways can give direction to pharmacological studies. The current chemotherapy drug is temozolomide (TMZ), an alkylating agent that is taken orally and readily penetrates the blood–brain barrier (114). It is often given at low doses concomitant with radiotherapy, followed by adjuvant standard doses. This aggressive treatment schedule results in an increase of the 2‐year survival rate from 10.4% with radiotherapy alone to 26.5% (150). Recent research has shed some light on why TMZ is not as effective as first thought. Shi et al has shown that microRNAs (small noncoding RNA molecules that regulate protein expression by cleaving or repressing the translation of target mRNAs), specifically mi‐RNA‐125b overexpression, might confer GSC resistance to TMZ as mi‐RNA‐125b is overexpressed in GSCs (141). Following the inhibition of mRNA‐125‐b in GSCs, the expression of the anti‐apoptotic protein Bcl‐2 was decreased, while the expression of the proapoptptic protein Bax and cytochrome c release from mitochondria was increased (141). Currently, there is no pharmacological cure for glioblastoma tumors and as surgical resection and radiotherapy are inadequate in being able to remove such a diffuse tumor, further developments in targeting GSC populations with chemotherapy treatments remain a priority.

Table 3.

Signaling cascades commonly found in GSCs and their role within the cell. Abbreviations: Akt = v‐akt murine thymoma viral oncogene; BAX = B‐cell lymphoma 2 associated X protein; BCL‐X_L = B‐cell lymphoma extra large; BCL = B‐cell lymphoma; BCL‐2 = B‐cell lymphoma 2; BMP = bone morphogenetic proteins; Bmi‐1 = B lymphoma Mo‐MLV insertion region 1 homolog; BMPR = bone morphogenetic protein receptor; CDK4 = cyclin‐dependant kinase 4; c‐Myc = myelocytomatosis; CSL = citrate synthase‐like; EGFR = epidermal growth factor receptor; Fra = fos related antigen; GSC = glioblastoma stem cell; Hes = hairy and enhancer of split 1; Hey = Hairy/enhancer‐of‐split related with YRPW; IGFBP‐2 = insulin‐like growth factor‐binding protein‐2; IGF = insulin‐like growth factor; MCL‐1 = myeloid cell leukemia‐1; PDGF‐β = platelet derived growth factor‐beta; NMDA = N‐Methyl‐D‐aspartic acid; NPC = neural precursor cell; NSC = neural stem cell; Rtk = receptor tyrosine kinase; IRS1 = insulin receptor substrate 1; WNT = Wingless‐type MMTV integration site family

Signaling pathway	Receptor(s)	Target gene(s)	Overall role of signaling pathway	Reference
Notch signaling	300 kDa single‐pass short‐range transmembrane receptor	Converted CSL becomes activator of transcription for Hes and Hey families, cyclin D and c‐myc	Essential for the maintenance and fate determination in somatic stem and progenitor cells by promoting self‐renewal and repressing differentiation	17, 78, 93, 123, 130, 139, 156
Stat3	Ciliary neurotrophic factor family cytokine receptor	BCL‐2 family including BCL‐2, BCL‐X, BAX and MLC‐1	Controls major cellular responses, including cell proliferation and survival. Also, the JAK‐STAT3 signaling pathway plays an essential role in the maintenance of NPCs by fibroblast growth factor 2.	21, 31, 54, 125
Rtk‐Akt	Receptor tyrosine kinases	EGFR	Activation of EGFR promotes the growth of both astrocyte precursors and neural stem cells.	3, 95, 144
Transforming growth factor beta	Serine/threonine kinase complex membrane receptor	PDGF‐β	Oncogenic stimulus in glioblastoma growth through induction of angiogenesis, immune evasion and invasion	20, 118, 173
Hedgehog‐gli signaling	Surface membrane receptor patched	Gli‐1, Cyclin D, Cyclin E, IRS1	Regulates the behavior of stem cells particularly clonogenicity, survival, tumorigenicity and proliferation	29, 37, 72, 88, 106, 132, 137, 175
BMP	BMPR1A; BMPR1B; BMPR2	Activin response element	Mediate a wide range of biological responses in NSCs including proliferation, differentiation and apoptosis. Instructive role in stem cell niche which favors the acquisition of the astroglial fate. BMP4 may inhibit GSCs by decreasing symmetric cell cycles that generate two GSCs on division, trigger differentiation or block proliferation	5, 98, 115, 120
Rac1‐Pak	NMDA	Cyclin D1	Maintenance of stemness and malignant properties of glioma stem‐like cells	83, 158, 178
Wnt‐β‐catenin	Low‐density lipoprotein receptor related protein 5, 6, 10; Frizzle receptor	Wnt 1, 5a, 10b, 13; Fra‐1; c‐Myc; Cyclin D1	Induces neuronal differentiation of neural precursor cells and promotes proliferation. It is also an important regulator of stem cell self‐renewal.	63, 71, 79, 84, 90
IGFBP‐2	IGF‐receptor	Cyclin D1, CDK4, Cyclin E, Bmi‐1,	GSC promoting factor and a broad regulator of GSC resistance to lethal stress induced by genotoxic agents or small molecule inhibitors. Stimulates the migration of stem cells	10, 73

In the RTK‐Akt signaling pathway, Soeda et al showed tyrosine kinase inhibitors (AG1478 and gefitinib) which are relatively selective for epidermal growth factor receptor (EGFR) kinase, eradicated the self‐renewal capacity of GSCs and the proliferation and differentiation of CD133+ cells (145). Eyler et al used a phosphatidylinositol ether lipid analog as an inhibitor of Akt and found that it decreased GSC migration and invasion and also increased the survival of immuno‐compromised mice bearing human glioma xenografts by 22 days (39). Thus, there is increasing evidence that by inhibiting a signaling pathway involved in GBM development, such as the Akt pathway, glioblastoma development can at least be hindered. Gallia et al also used an inhibitor of the Akt pathway, the small polymer molecule A‐443654, finding that it inhibited the growth of GSC lines and extended the survival of intracranial glioma tumor bearing mice by an average of 11 days when they received the tumor and polymer on the same day and an average of 6 days when they received the polymer 4 days after receiving the tumor (50).

More recently, Griffero et al evaluated the in vitro sensitivity of human glioma tumor initiating cells to (EGFR) kinase inhibitors and found that they are responsive to anti‐EGFR drugs erlotinib and gefitinib, but phosphatase and tensin homolog expression and Akt inhibition seem to be necessary for such effect (58). An atypical approach of using eckol, a phlorotannin component of some brown algae, has been used to suppress the stemness and the malignance of glioma stem‐like cell populations (75). Hyun et al found that eckol inhibited the phosphoinositide 3‐kinase‐Akt and Ras‐RAF‐1Erk signaling pathways, which also made glioma stem‐like cells more sensitive to anticancer treatment such as ionizing radiation and temozolomide (75).

Research has shown that CXCR4 can signal through the Akt and RAS pathways 117, 163, 174. Redjal et al used the CXCR4 inhibitor AMD3100 in combination with BCNU on GBM cell lines and found it resulted in synergistic antitumor efficacy in vitro and also resulted in tumor regression in vivo showing increased apoptosis and decreased proliferation of mice with established orthograft tumors (127). Furthermore, CXCR4 is the known receptor for stromal cell derived factor‐1 and has also been shown to mediate neural stem cell migration toward injury or tumors in a murine brain (135). Schulte et al confirmed CXCR4 overexpression in GSC lines by real‐time polymerase chain reaction analysis, and then used a CXCR4 receptor blockade, AMD3100, for in vivo treatment of glioblastoma growth in the brains of nude mice and found that it caused inhibition of expansive tumor growth by 58.3% (135).

It has been suggested that Notch may play a particularly important role in cancer stem cells and it has been shown to be critically implicated in stem cell fate determination and cancer in GSCs 121, 122, 142. Notch proteins are transmembrane receptors that mediate short‐range cellular communication through interaction with ligands (Jagged‐1, ‐2 and Delta‐like‐1, ‐2, ‐3, ‐4). The activation of Notch requires sequential proteolytic cleavage by the gamma secretase complex to release its intracellular domain from membrane to nucleus. Notch signaling promotes the proliferation of normal neural stem cells and is required for the maintenance of neural progenitors both in vitro and in vivo (108). A study by Wang et al showed that inhibition of the Notch pathway with gamma‐secretase inhibitors (GSIs) renders the GSCs more sensitive to radiation at clinically relevant doses (170). Indeed, inhibition of the Notch signaling pathway with GSIs has been the most extensively used approach to target Notch (93). However, it has been noted that GSIs affect all Notch receptors, causing unwanted side effects, the most common being metaplasia of goblet cells in the small intestine (147).

A study by Li et al of U87 human glioblastoma cells and glioma stem cells 0308 interestingly found that microRNA‐34a (miRNA‐34a), a transcriptional target of p53, down‐regulated notch‐1 and notch‐2 expression in glioma cells (97). The transient transfection of miRNA‐34a in glioma stem cell lines strongly inhibited cell proliferation, cell cycle progression, cell survival and cell invasion, showing potential as a therapeutic agent for GSCs (97). It should be noted that miRNA‐34a has multiple predicted targets among which are many oncogenes, and therefore the proposed inhibition of malignant activity may not have been strictly caused by the down‐regulation of notch‐1 and notch‐2. More recently, Campos et al used all‐trans‐retinoic acid (ATRA), a known potent modulator of cellular differentiation and proliferation, as a treatment on stem‐like glioma cells and found that expression of Notch‐1 was reduced on mRNA and protein levels after differentiation (23). Following implantation into mouse brains with undifferentiated stem‐like glioma cells, there was higher Notch‐1 expression as compared with samples which were inoculated with differentiated stem‐like glioma cells (23). On post‐mortem examination of the brains of the mice, it was reported that those which had been implanted with undifferentiated cells developed tumors which were more than seven times larger and showing a highly invasive growth pattern, compared with those mice which were implanted with differentiated cells, despite being implanted with 10 times fewer cells (23).

Further work on the use of ATRA with GSC has been carried out by Ying et al. The group used ATRA treatment to induce changes in morphology, induce growth arrest at G1/G0 to S transition, decrease cyclin D1 expression and increase p27 expression in GSCs which were grown as nonadherent neurospheres in growth factor supplemented serum‐free medium (177). As with Campos et al, it was also found that several Notch pathway components were significantly down‐regulated by ATRA (177). Furthermore, the importance of the Notch signaling pathway cannot be underestimated, as indicated by a study by Wang et al which showed that either notch1 or notch2 protected GSCs against radiation (170), indicating that currently, attaining the full benefits of radiotherapy for GBM treatment is only likely to occur once significant advances have been made pharmacologically.

Clement et al tested for the possible role of hedgehog‐gli signaling in cancer stem cells, using glioma stem cell cultures, which self‐renew and mimic the original tumor after transplantation into the brains of immunocompromised mice (29). Treatment with cyclopamine decreased their proliferation and increased apoptosis in a concentration‐dependent manner (29). It was also argued that due to the lack of any obvious secondary effects in cyclopamine‐treated adult mice, it was suggested that such treatment may spare normal quiescent stem cells in their niches, likely allowing regeneration of any damage to normal adult tissues after cessation of treatment (29). Similarly, Bar et al found that cyclopamine caused a 40%–60% reduction in growth of adherent glioma lines which largely expressed hedgehog‐gli signaling (8). When the GBM‐derived neurospheres were treated with cyclopamine and then dissociated and seeded in media lacking the inhibitor, no new neurospheres formed, suggesting that clonogenic cancer stem cells had been depleted (8). Consistent with this hypothesis, the stem‐like fraction in gliomas marked by both aldehyde dehydrogenase activity and Hoechst dye excretion was significantly reduced or eliminated by cyclopamine (8). More recently, Ulasov et al found that the therapeutic effect of TMZ was enhanced by Notch and Hedgehog pathway pharmacological antagonism with GSi‐1 and cyclopamine (162). Importantly, simultaneous treatment involving TMZ with both GSi‐1 and cyclopamine led to a significant increase in CD133+ glioma cytotoxicity than treatment with any of these agents alone (162).

Other notable pharmacological developments include the use of GSC lines derived from right temporal GBMs, in which Sherry et al found that inhibition of STAT3 in these cells with either small molecular inhibitors or RNAi resulted in inhibition of growth and neurosphere formation although they did not undergo apoptosis (140). Wei et al also showed that STAT3 expression in GSCs can be inhibited with a STAT3 short hairpin RNA or the small molecule inhibitor WP1066 (172). More recently, Ma et al investigated the role of celecoxib, a selective COX‐2 inhibitor, in enhancing the therapeutic effects of radiation on CD133+ glioblastoma cells (103). They found that high levels of COX‐2 protein were detected in CD133+ but not in the CD133− glioblastoma cells and further demonstrated that 30 µM celecoxib was able to enhance the effect of radiation in inhibiting colony formation and increasing apoptosis in CD133+ glioblastoma cells (103).

Despite recent advances in novel pharmacological developments aimed at the various signaling pathways which exist in GSCs, one of the major difficulties in developing any pharmacological treatment for GBM is that cell signaling cascades differ depending on the GBM and it is clear from the studies that not all signaling pathways are active or required to produce GBM. Indeed, in certain signaling cascades, there is still debate regarding their existence in grade IV GBM tumors. The hedgehog‐gli signaling pathway, for example, has been shown to be active in more than 50% of GBM (8). When Bar et al used Gli‐1 as a marker for the hedgehog pathway, it was shown to be highly expressed in five out of nine primary GBM and in four of seven GBM cell lines. However, other research demonstrated in 13 adult gliomas that hedgehog‐gli pathway was operational and active within grades II and III gliomas, but not grade IV GBM (137). It is likely that hedgehog signaling is operational in only some brain tumors, as demonstrated by Xu et al who showed there are both hedgehog‐gli signaling‐dependent and ‐independent brain tumors characterized by a cancer stem cell zone of Gli‐1 positive cells (175). This indicates that within any given GBM tumor, there are most likely a number of different types of GSCs operating, each with its own unique and multivariant signaling cascades.

GENE MUTATIONS IN GSCs

The genetic research of GSCs, which can be seen in Table 4, clearly shows that there are a number of mutations which are involved in the development of brain tumors. So far there is no evidence that any one individual gene can induce glioblastoma tumor formation. Studies have shown that the activation of certain genes may also require the specific activation of another gene, combinational gene activation, in order to induce glioblastoma formation 69, 105. This evidence highlights the importance of combined gene mutations which have to occur synergistically in order to achieve a desired effect. It is possible that many genes and signaling pathways require similar co‐operators in order for their effect to take place and therefore a more holistic observation of cellular activity may be required in order to better understand genetic and signaling interaction.

Table 4.

Gene mutations, mechanism of action and involvement in glioblastoma development. Abbreviations: Akt = v‐akt murine thymoma viral oncogenes; Arf = ADP‐ribosylation factor; Bmi‐1 = B lymphoma Mo‐MLV insertion region 1 homolog; ECM = extra cellular matrix; EGFR = epidermal growth factor receptor; GBM = glioblastoma multiforme; mTOR = mammalian target of rapamycin; NPC = neural precursor cell; NSC = neural stem cell; OPC = oligodendrocyte precursor cell; PDGF‐β = platelet derived growth factor‐beta; PTEN = phosphatase and tensin homolog; TGF‐β = transforming growth factor beta; Ras = RAt Sarcoma

Gene	Action in neural stem cell	Involvement in glioblastoma development	Reference
p53	Negative regulation of NSC proliferation. Mediator of cellular responses to DNA damage, stimulating and increasing the levels of Cdk inhibitor Cdkn1A	Deficiency induces high levels of platelet derived growth factor receptor expression, deregulation of the Rb‐mediated cell cycle regulatory pathway, high levels of Cdk4 expression and cyclin D1 over expression.	99, 116, 171
Ras and Akt	Ras gene codes for proteins whose primary role is to assemble transient signaling complexes at the membrane that activate signal transduction pathways coordinating transcription, cell shape and migration, endocytosis, cell survival and cell cycle progression. Akt is a downstream target for receptor tyrosine kinases that are regulated by stimuli such as growth factors and insulin, and therefore plays a major role in metabolism, cell growth and cell survival.	Ras, if mutated prevents the conversion of Ras‐GTP back to Ras‐GDP, stimulating continuous cell growth due to the consistent stimulation of a signal transduction pathway for cell proliferation. Akt activity in tumors correlated with phosphorylation of S6RP and 4E‐BP1 downstream of mTOR. Transfer of genes encoding for Ras and Akt together to neural progenitors induces high grade gliomas.	15, 53, 80, 105, 116
p16INK4a and p19ARF	p16INK4a inhibits cyclin D‐dependent kinases, regulates the ability of the retinoblastoma protein to control G1 exit. p19ARF is a negative regulator of Mdm2 function, can activate p53 in response to particular oncogenic signals, thereby inducing cell cycle arrest or apoptosis, depending on the biological context.	Loss of both p16INK4a and p19ARF in combination with epidermal growth factor receptor pathway activation provokes a common high‐grade glioma phenotype regardless of special cell or origin.	85, 95
Bmi‐1 transcription factor	A core component of the Polycomb group proteins which are epigenetic chromatin modifiers involved in heritable gene repression and maintenance of stem cell self‐renewal and progenitor proliferation and this has been primarily attributed to its ability to repress the Ink4a‐Arf locus	Bmi‐1 inactivation causes a change in the expression of genes involved in the TGF‐β signal transduction pathway, ECM remodeling and cell cycle control. Bmi‐1 can cause tumor development in an Ink4a/Arf‐independent manner.	18, 19, 109, 151,
PDGF‐β	Exerts its stimulatory effects on cell growth and motility by binding to two related protein tyrosine kinase receptors	Deregulated PDGF‐β expression has been shown to cause glioma‐like tumors by retroviral insertion into newborn mice.	66, 110
PTEN phosphatase	Negative control of NSC proliferation.	Frequently mutated in aggressive high‐grade gliomas.	59, 131, 161
EGFR membrane receptor	EGF‐dependent NSC proliferation/maintenance	Transduction of Ink/Arf‐/‐ in NSC or astrocytes with constitutively active EGFR induces high‐grade glioma in mice. Enhance cell proliferation by promoting P13K/Akt signaling; inhibit cell cycle regulators including 27 KIP1 and increase expression of anti‐apoptotic proteins such as Bcl‐XL and unregulates vascular endothelial growth factor, interleukin‐8 and matrix metalloproteinase 13 expression	51, 95, 148, 160
High‐mobility group A1	Developmental regulator of stem cell self renewal and Ink4a/Arf expression	Important role in the proliferation, invasion and differentiation of GBM	40, 111

Indeed there is growing evidence that there is no single universal GSC type. Phillips et al reported classification of three main glioblastoma subtypes defined by gene expression signature. Three subtypes included one which expressed neurodevelopmental genes which was associated with a better prognosis, and two other subtypes characterized by their resemblance to either highly proliferative cell lines or tissues of mesenchymal origin, both being associated with a rapid rate of cell division (119). Similarly, Guenther et al established glioblastoma stem‐like cell lines from nine different glioblastomas which had been maintained for up to 3 years and although established under identical conditions two distinct subtypes arose. Using gene expression profiling and considering neurobiological criteria, four cell lines (cluster 1) shared similar gene expression patterns associated with neural development and displayed a full stem‐like phenotype, whereas five cell lines (cluster 2) shared a different gene expression pattern and displayed only restricted stem‐like phenotype (62). Interestingly, most of the cluster 2 cell lines showed no detectable CD133+ cells, although these cell lines did contain cells expressing the Sox2 and nestin markers. More recently, a study by Lottaz et al compared the gene expression profiles of 17 GSC lines and distinguished two subgroups; type I GSC lines display “proneural” signature genes and resemble fetal neural stem cells lines, whereas type II lines show “mesenchymal” transcriptional profiles similar to adult neural stem cells (100). Clearly then, GSCs are not represented by one single phenotype or genetic signature but rather a mosaic, which may be typical of the environment in which they develop.

Furthermore, it has been shown that specific combinations of gene deletions lead to different tumors. Bachoo et al found that combined loss of p16^INK4a and p19^ARF, but not of p53, p16^INK4a or p19^ARF, enables astrocyte dedifferentiation in response to EGFR activation (3). Similarly, Jacques et al deleted Rb/p53, Rb/p53/PTEN or PTEN/p53 in adult subventricular stems and in ectopically neurografted stem cells (77). It was found that recombination of PTEN/p53 gave rise to gliomas whereas deletion of Rb/p53 or Rb/p53/PTEN generated primitive neuroectodermal tumors. It would seem, therefore, that stem cells which have activated oncogenes and inactivated tumor suppressor genes give rise to tumors of specific phenotypes depending on the combination of mutations.

RELATIONSHIP BETWEEN IMMUNOSUPPRESSION AND ONCOGENESIS RELATIVE TO GLIOBLASTOMAS

A relationship must exist between immunosuppression and oncogenesis, and for GSCs it might be that they sustain neoplastic growth through immunoevasion and/or some kind of immunomodulatory function. This correlates with the fact that many immunocompromised patients are at a higher risk of developing cancer (60). GSCs can disrupt tumour immunosurveillance and result in both ineffective adaptive and innate immune responses (55). There are a plethora of immunosuppressive mechanisms associated with malignant gliomas, where patients have demonstrated significant impaired immune function (179). These mechanisms include inhibition of antigen‐presenting cells, direct inhibition of B cells and T cells, and the immunosuppressive effects have been shown to be mediated by chemokines such as transforming growth factor beta, prostaglandin E₂ and certain gangliosides (36). A recent report by Rodrigues et al (129) also found that GSCs suppress activation of microglial cells, which in turn suppress T‐cell activity by two secreting immunosuppressive factors, interleukin‐10 and Fas‐ligand. Thus, GSCs contribute to immunosuppression, and given the poor prognoses associated with high‐grade gliomas, there is a therapeutic need in order to improve survival rates. Owing to the fact that the immune system must play a key role in the development of GBM, it would be highly desirable to develop an immunotherapeutic treatment strategy. More importantly, it may be necessary to take a multifaceted approach to treatment whereby two or more immunotherapeutic strategies are adopted, and this approach will be more beneficial to GSCs due to the inherent heterogeneous population of cells with different proliferative potential.

CONCLUSION

The continual growth in research and new findings on glioblastomas and the causative subpopulation of stem cells highlight both its importance and the challenge it presents. This type of cancer introduces a number of unique problems, including locating the origin of stem cells which become tumorigenic not least because of a lack of a global cell marker for this type of tumor. The variability in signaling pathways and genetic mutations would indicate that at least a few, if not many, of these are active within a single GSC population, indicating that any pharmacological treatment will have to be broad enough to cover all active tumorigenic signaling pathways and mutations in order to be successful. Thus, the evidence seems to propose that glioblastomas are derived from multiple stem cell lineages, whereby there are subpopulations within the subpopulation of GSCs. This would help to explain the variety of combinations and permutations of cell markers, signaling cascades, and genetic mutations found within single tumor samples and across all samples from different patients. Furthermore, the nutrient‐rich brain matter provides an optimum environment that allows for the extraordinary proliferative capacity of GSCs to develop, meaning an early diagnostic marker is crucial. Future studies may want to focus on identifying any differences between cancer stem cells which make up the subpopulation of a glioblastoma and consider if it is possible to provide treatments which are multidimensional in their approach.