Abstract
This review begins with the description of the nervous cytogenesis, proliferation of primitive cells, migration and differentiation with antigen expression and regulation through factors. Emphasis is given to neural stem cells, progenitors and to radial glia as belonging to early stages of gliogenesis. Experimental production of brain tumors in the rat by transplacental administration of ethylnitrosourea and systemic administration of methylnitrosourea effective through neural stem cells and progenitors is briefly described as a preamble to the recent conception of gliomas as originating from stem cells or brain tumor stem cells (BTSCs). The nature and origin of BTSCs, their molecular regulation and their recognition in vivo are discussed. In the growth of tumors, the role of nestin, migration of neural stem cells toward the tumor from the subventricular zone or from the tumor to the brain and the migration from bone marrow of mesenchymal stem cells for angiogenesis of the tumors are considered. Special mention is made of the relationship between glio-vascular niches in the subventricular zone and the neo-niches around the tumors with their importance in the tumor growth as points of joint between cytogenesis and angiogenesis. Finally, besides the importance of hypoxia, cell traffic in the brain adjacent to tumor is emphasized in relation to the growth of gliomas.
The cell origin of brain tumors has been of the greatest interest since the first histogenetic classifications. The relationship between the morphology of glioma cells and that of cells during cytogenesis, from primitive neuroepithelial cells to their derivatives, was at the basis of Bailey and Cushing's classification (1). However, a number of in vivo and in vitro observations have progressively shaken the reliability of such a relationship for diagnosis and prognosis.
The current main hypothesis on the origin of cancer, and also for gliomas, is based on the somatic mutation theory (2), even though the opposing theory of field conversion (3) has not been completely forgotten. The cell origin of gliomas is still a matter of discussion, and the stem cell theory has recently gained special attention, especially for its possible therapeutic implications. The concept of stem cells is rather vague if referring to a specific cell type and the concept of ‘stem-like status’ was introduced to encompass not only progenitors and precursors, but also tumor cells showing certain stemness properties. A global view of the problem generated the aphorism ‘it is not that glioblastoma is a tumor which affects the brain, but that it is the diseased brain which expresses glioblastoma’. In the growth of gliomas and in their interaction with the hosting organ, the brain adjacent to tumor (BAT) acquired a special interest as the place where intense traffic of cells and factors occurs.
Nervous Cytogenesis and Neural Stem Cells
In humans, it is still impossible to identify the cells of origin of the tumors prior to their transformation (4), because the earliest stages of glial tumor development are not known and the first visible lesions, for example astrocytomas, are already organized as tumors when they become recognizable. Neurons and glia cells derive from self-renewing and multipotent primitive neuroepithelial cells of germinative zones (Figures 1 and 2). They differentiate along different pathways under extrinsic and intrinsic stimulations. Neurogenesis occurs first through organizing centers which generate signals inducing the expression of patterning genes, encoding transcriptional factors and controlling neuronal subtypes in the adjacent neuroepithelial cells (5). Gliogenesis follows after neurogenesis has been completed. Neuroepithelial cells migrate from the germinative zone (VZ) (Figure 1A, B), which later forms the subventricular zone (SVZ) (Figure 1C), proliferating and differentiating into glial cells. The differentiation proceeds through progenitors and precursors defined by morphology, expression of markers and ability to differentiate into specific cell types in culture (6) (Figure 2A).
Primitive neuroepithelial cells are supposedly neural stem cells (NSCs) because they are pluripotent, self-renewing, proliferate and express a series of antigens such as nestin, CD133/prominin-1 (7), musashi-1 (8), sex determining region of Y chromosome (Sry)-related high mobility group box2 (SOX2) (9) etc. In the course of cytogenesis, antigens are expressed in temporal order: nestin, vimentin, neuron cell surface antigen (A2B5), glial fibrillary acid protein (GFAP) and then O4, proteolipid protein, galactocerebroside, myelin basic protein, synaptophysin and neurofilaments (Figure 2A, B). Growth factor signaling controls the passage from one stage to the other: platelet-derived growth factor (PDGF) and beta-fibroblastic growth factor (bFGF) promote the passage from stem cells to precursors and PDGF to O2A progenitors, whereas the passage to astrocyte precursors is promoted by PDGF, cliary neurotrophic factor (CNTF) and epidermal growth factor (EGF). The passage from O2A progenitors to oligodendrocytes and type 2 astrocytes is promoted by CNTF and EGF and inhibited by PDGF and bFGF (4, 10). During migration, glia cells continue to proliferate (Figure 1C). At the end of the process, the SVZ decreases in size, persisting in the adult as sub-ependymal cell layer (Figure 1D).
NSCs can be defined as multipotent and capable of giving rise to the cells of mature brain. Under the influence of growth factors EGF and bFGF (11), they form neurospheres that can be dissociated and replated with an exponential increase of their number (12). In the adult, NSCs are found in specialized niches such as the dentate gyrus of hippocampus (SGZ) and the SVZ where different cell types occur (Figure 2D): A cells are neuroblasts migrating to the olfactory bulb; B cells are NSCs with morphological features of astrocytes and express receptors for and respond to PDGF and EGF. Mitogen-treated, B cells give rise to transit-amplifying type C cells, which are precursors of neuroblasts and type A cells, positive for poly-sialated neural cell adhesion molecule (PSA-NCAM), β-III tubulin and nestin, which migrate to the olfactory bulb and then mature into neurons (13). C Cells are positive for nestin, whereas B cells are positive for GFAP, vimentin and nestin and would represent the real NSCs (14), also generating oligodendrocytes. PDGF functions as a mitogen for the precursors of the latter (15).
The stem cell-vascular niche complex also contains endothelial and mural progenitor cells (16) (Figure 2D). In the SGZ, neurogenesis occurs in foci closely associated with blood vessels (17). In serum-containing medium, normal NSCs tend to differentiate spontaneously as a monolayer into neurons, as shown by β-tubulin and neuronal specific enolase (NSE) and glia cells, as shown by glial fibrillary acidic protein (GFAP) and O4. This could mean that neurospheres from SVZ cells, both in embryos and in the adult, are ultimately astrocytes; radial glia also may represent stem cells or progenitors during embryonal development (18), in line with the hypothesis that stem cells are found within the lineage of neuroepithelial cells – radial glia – astrocytes (15, 19). This hypothesis includes the possibility that glia cells, or a subset of them, in the adult may represent latent stem cells throughout the brain and there is evidence that cells from non-neurogenetic regions, if cultured with bFGF/EGF, become neurons (20). Both SVZ and SGZ contain astrocytes and since adult SVZ stem cells may derive from radial glia that in turn derives from the primitive neuroepithelium and a subset of cells expressing GFAP are putative adult NSCs, there would be a continuum from neuroepithelium to radial glia to astrocytes, considered as the astrocytic lineage, capable of producing NSCs (21).
Neurogenesis and gliogenesis continue in adult mammalian brain (22, 23) and the existence of NSCs in the adult is a basic working concept for the origin of tumors. The role of the oligodendrocyte transcription factor 2 (Olig2), a helix-loop-helix transcription factor, which is expressed in gliogenic progenitors in the postnatal SVZ (24) is also important. It takes part in the cell cycle control of progenitors and it is expressed in those giving rise to oligodendrocytes and certain neuron subtypes (25, 26), i.e. in NG2 cells capable of forming oligodendrocytes or astrocytes (27). Olig2 is considered a kind of gatekeeper for brain tumor stem cells (BTSCs) of which it regulates self-renewal and multipotency. Most astrocytoma cells and all CD133+ cells are positive for Olig2 (28).
Radial Glia
Radial glia derives from neuroepithelial cells at the onset of neurogenesis; the soma borders the ventricle and the processes extend to the pial surface as scaffolding to migrating neurons (Figure 2C). Radial glia has the features of astrocytic lineage and most progenitors possess radial glia features. It expresses RC2, nestin, vimentin, GFAP, glutamate/aspartate transporter (GLAST) and it is neurogenetic (29). Radial glia state is maintained by notch signaling through ligand delta1, by ErbB signaling through neuroregulin and by FGFR signaling through FGF (29), which on the other hand are known to be critical for glioma cell survival and proliferation (30). For the establishment of radial glia processes, a glycoprotein secreted by cells in the marginal zone, reelin (31), is very important. At the end of neuron migration, radial glia transforms into astrocytes, as committed to the astroglial lineage (32) and as a form of immature glia. It reveals stem cell characteristics as do the derived astrocytes, either during development or in the adult (20, 33). Radial glia cells may function as progenitors, in line with the hypothesis that NSCs in the adult brain would be contained within the astrocytic lineage as a continuum (15, 18, 19, 21, 34). In the neonatal ventricular wall of mice, radial glia produces astrocytes, oligodendrocytes, ependymal cells and neurons.
Experimental Brain Tumors Induced by Nitrosourea Derivatives
Our knowledge on tumor arising from stem cells of VZ, SVZ, or SGZ during embryonic or in adult life is mostly based on experimental studies of glioma induction in the rat by nitrosourea derivatives. Ethylnitrosourea (ENU) is active in neonates and transplacentally, and methylnitrosourea (MNU) if administered into the brain in adults. The tumors produced originated from the VZ and from the derived SVZ (35-39) by the alkylation of O6 of guanine, O2 of cytosine and O2 and O4 of thymine, followed by coupling errors during transcription (40) and defective repair of DNA (41). Short-term effects were mitotic arrest and nuclear death in the germinative zones (42), and apoptosis of neural precursor cells with caspase-3 activation, largely dependent on p53 and caspase-9 (43). Long-term phenotypic effects were early lesions, microtumors and tumors in the brain hemispheres, starting in the periventricular white matter (44) (Figure 1E, F, G, H; Figure 2E).
Various tumor types arose on the basis of the concept of a ‘window of vulnerability’ for each precursor cell type (45, 46). The concept later emerged that, experimentally, the more advanced the differentiation status of progenitor cells is, the greater genetic alterations must be for tumor transformation (47). It was then demonstrated that cells of the SVZ or of the early lesions in the white matter were nestin positive, as were cells cultivated from SVZ of exposed rats, confirming that tumors develop from stem or precursor cells (48).
MNU administered intracerebrally, intraperitoneally or subcutaneously in the adult, when the germinative zone is no longer present, gave rise to tumors in the periventricular white matter, corpus callosum and hippocampus (36), originating from the SVZ or the sub-ependymal layer.
BTSCs
Comparing neuroepithelial cells during cytogenesis and glioma development, the existence of characteristics common to progenitor cells and malignant is apparent: simplicity of form, proliferation capacity, potential to differentiate and capacity to migrate (49). Moreover, some of the genetic alterations that characterize malignant gliomas, such as the activation of signal transduction pathways and disruption of the cell cycle machinery, which represent the molecular signature of these tumors, concern genes and proteins involved in the regulation of differentiation during cytogenesis and may influence the differentiation/dedifferentiation status of the cells (49). In the differentiation process, EGF, FGF, PDGF, CNTF, insulin-like growth factor (IGF), sonic hedgehog homolog (SHH) etc. are mitogenic for cell proliferation and EGF/EGFR, signaling to Ras/mitogen-activated protein kinase (MAPK), influences the number of astrocytes or extent of apoptosis (50). CNTF promotes astrocytic differentiation (51) that, on the contrary, can be prevented by CpG methylation of signal transducer and activator of transcription 3 (STAT3) (52). Notch regulates astrocytic and oligodendrocytic differentiation and PDGF is involved in that of oligodendrocytes etc. (47). In mice, PDGFRα+ B cells generate new neurons and oligodendrocytes, and excessive PDGF activation in SVZ induces early stages of tumor formation (53). Infecting adult rat white matter with retrovirally introduced PDGF drives supposed glial progenitors to form malignant gliomas (54).
In humans, the possibility to trace back the origin of the tumors from the aspect of their cells is impossible. The morphology does not entirely correspond to that of mature cells of a certain line, but it is rather given by a mixture of cells resembling stages of cytogenesis, or even expressing markers of immature cell types (49). As a consequence, the tumor cells will appear dedifferentiated if compared with mature cells or, alternatively, they seem to originate from progenitor-like cells in the tumor. Basically, after a cell has received a transformation it may undergo differentiation or dedifferentiation (55).
The resemblance of tumor cells to those formed through cytogenesis is an unreliable method for establishing the cells of origin of gliomas, because differentiation during cytogenesis may undergo environmental, epigenetic and genetic influences. The histology of a tumor ‘would be more a reflection of the environment and time of initiation than the cell of origin’ and this would decide whether a tumor ultimately becomes, for example, an astrocytoma or an oligoastrocytoma (48). Tumor cell heterogeneity is due in part to epigenetic variations in progenitor cells (56). Introducing protein kinase B (Akt) and K-Ras in mouse brains by a retroviral technique, it is easier to obtain tumors in nestin-expressing than in GFAP-expressing cells, especially if there is a loss of cyclin-dependent kinase inhibitor 2A (CDKN2A) (57). Therefore, genetic deregulations would appear to be more important than the cell stage of origin.
Epigenetic events could be responsible for dedifferentiation: in U373 MG cell lines of glioblastoma, transforming growth factor α (TGFα) or other factors acting on tyrosine kinase receptor protein (TKRP) can reduce GFAP mRNA and enhance nestin expression, affecting or not vimentin (58, 59). The transforming event could also block the differentiation of NSCs (60), according to the old concept of maturation arrest (61), and it could affect either an NSC or a tumor cell which has re-acquired properties of an NSC. Other examples are available: in rat progenitor cells, the overexpression of Akt or K-Ras produces tumors with the phenotype of human glioblastomas, in which, on the other hand, they are overexpressed (62), whereas the overexpression of PDGFRβ produces tumors with the phenotype of oligodendrogliomas (63). Other studies indicate Myc as having a role in the formation of gliomas, promoting or reinforcing an undifferentiated phenotype required for glioma cells to respond to the oncogenic effects of elevated K-Ras and Akt activity (64). Different phenotypes may have the same genetic basis (65). Practically, the neoplastic transformation of glial precursors produces tumors with the phenotypes of astrocytoma and oligodendroglioma and this depends on the activation or inactivation of specific protein pathways. A greater knowledge of the relationship between molecular pathways and tumor phenotype is needed for clarifying the origins of gliomas that at the moment are tentatively deduced only from the phenotype of tumor cells.
The hypothesis that cancer stem cells (CSCs), and then BTSCs, exist is based on the concept that cancer growth, clonal diversification and evolution, metastasis and recurrence after therapy are due to a small fraction of cells possessing the capacity for infinite proliferation, self-renewal and tumorigenicity (66, 67). BTSCs have been isolated from benign and malignant gliomas and are believed to derive from transformed NSCs, because both cell types share the same specific markers, i.e. nestin, musashi-1 and CD133. The basic concept is that preservation of markers in daughter cells indicates self-renewal, and expression of markers of adult neurons and glia indicates differentiation (68, 69). Self-renewal is demonstrated by the passaging assay, i.e. replating cells dissociated from neurospheres under serum-free medium conditions.
The properties of NSCs, i.e. lack of differentiation, capacity to form neurospheres (Figure 3A) and to proliferate, self-maintenance, clonogenicity (11) and positivity for a series of markers, can be recognized in BTSCs, but there are dissimilarities. In tumors, only a proportion of cells are clonogenic when xenografted (48) and this corresponds to the rule that in a tumor, the number of cells undergoing self-renewal is small and that a larger cell population differentiates (70). Of course, it is not easy to identify BTSCs in vivo and not all doubts have been dispelled on this matter (71). In vitro, cells generating clusters of clonally derived cells resembling neurospheres, which are self-renewing, proliferating and capable of differentiation have been demonstrated from brain tumors (66, 68, 69, 72). Transplanted into mice, they reproduce tumors with the characteristics of human glioblastoma (69) (Figure 3B). They have a higher growth rate and self-renewal capacity than NSCs, correlating with malignancy, as for growth rate and telomerase activity. BTSC neurospheres express neural stem cell markers, CD133 and nestin, and lineage markers, and differ from NSCs (68): they invade normal neurospheres, differentiate faster than NSCs and produce aberrant cells expressing glial and neuronal antigens, and not neurons and glial cells (73).
The expression of a series of antigens, among which musashi-1, CD133 and nestin, plays a great role in the recognition of BTSCs (Figure 3C, D, E). CD133 is a 120 kDa transmembrane glycoprotein, expressed on primitive cell populations and occurring in different types of tumors (7). It has been differently interpreted: initially, it was considered as the only indicator of BTSCs. Only a CD133+ cell subpopulation from human tumors would exhibit stem cell properties in vitro, capacity for self-renewal and reproduce tumors after transplantation (Figure 3B), whereas CD133− cells engrafted would not reproduce tumors (74). It was then demonstrated that CD133− cell populations also contain BTSCs (75-79). They differ from CD133+ cells regarding angiogenesis and radiotherapy-resistance, but not in their ability to initiate tumors (80, 81). CD133+ cells have also been found in perivascular niches of gliomas (82) and have been associated with unfavourable progression-free survival in malignant glioma (83, 84) and oligodendroglioma (85). CD133− cells not only show phenotypic hallmarks of BTSCs (77), but they can also give rise to CD133+ cells (77, 86). Positivity or negativity for CD133 can be influenced by the culture conditions and if isolated immediately after surgery, the CD133+ population does not coincide with that for CSCs. There is no difference between CD133-positive and -negative cells as regards clonogenicity and differentiation, telomerase activity by in situ hybridization, in pro-angiogenic activity, nor in repair activity of DNA after radiotherapy, in chemosensitivity and in the expression profile after comparative genomic hybridization (CGH). Of course, among these properties, clonogenicity is the most important as characterizing BTSCs (87).
Spheres derived from 9L gliosarcoma with EGFR and bFGF express nestin and SOX2 (Figure 3F), self-renew and differentiate into neuron-like and glial cells in vitro. They can propagate and recapitulate tumors when implanted into the brain of syngenic Fisher rats, displaying a more aggressive course compared with 9L gliosarcoma cells grown in monolayer cultures and devoid of mitogens. The sphere-generated cells have a lower proliferation rate, are more chemoresistant than 9L gliosarcoma cells grown as a monolayer, and express several antiapoptosis and drug-related genes, which may have clinical implications (88).
Tumor-initiating cells, besides glioblastoma and medulloblastoma, are also purported to exist in ependymomas (89). CD133+ cells showing features of radial glia can be cultured from ependymomas by neurosphere assay (90) and radial glia-like cells, expressing CD133, nestin, brain-lipid-binding protein (BLBP) and RC2, generate tumors in vivo. Genetic abnormalities may transform radial glia cells in cancer stem cells of posterior fossa and spinal ependymomas (91).
It is still debated whether BTSCs derive from the transformation of normal NSCs (47) or represent dedifferentiated cells through anaplasia, i.e. new clones developed by accumulation of mutations and selected by competition, with a high proliferation rate and lacking any differentiation antigen. Dedifferentiated cells might acquire stem cell properties. Paraffin fluorescent in situ hybridization (FISH) studies on CD133+ xenografts from a glioblastoma multiforme showed EGFR amplification, so that both CD133+ and CD133− cells bear the same cytogenetic alterations and therefore they are clonally derived (74). To know that BTSCs undergo asymmetric division, i.e. one daughter cell remains as a CSC in the VZ and the other migrates away and proliferates, does not help much, even though it is important for therapeutic consequences (92).
Four-seven independent mutations are necessary before a cancer phenotype appears in normal somatic cells (93) and some of these alterations, if affecting cells during development, may predispose to cancer (94). Progenitor, transiently dividing cells, may accumulate mutations and transform into tumor cells during development or in the adult SVZ, where a degree of cell proliferation is preserved (95). Three explanations have been proposed for the occurrence of cells positive for stem cell markers in brain tumors (96): they are (i) transformed NSCs expressing nestin and musashi-1; (ii) cells re-expressing nestin (97, 98) because they are dedifferentiated; (iii) exogeneous stem cells attracted to the tumor (99).
Cells from primary glioblastoma multiforme cultured in NBE (serum-free neurobasal media supplemented with basic FGF and EGF) conditions resemble NSCs as they form neurospheres, show potential for self-renewal and ability to differentiate, and show gene expression profiles similar to those of NSCs, but at the same time they show genetic alterations of primary tumors. If tumor cells are cultured in standard medium, they lose all these properties, including clonogenicity and tumorigenicity (100).
Cultures of NSCs from human fetal brain tissue and BTSCs from human glioblastomas have been compared in vitro. The former differentiate into mature phenotypes, whereas the latter grow flattened and attached for a week, then aggregate and reform spheres. CD133 and nestin expressions decrease in the first week and thereafter they increase. This indicates that BTSCs resist differentiation (101). Whereas secondary glioblastomas do not seem to contain BTSCs, primary glioblastomas contain CD133+ BTSCs which form neurospheres and show non-adherent growth. From them, CD133− cells also fulfill stem cell criteria and both are tumorigenic, but CD133+ tumors grow faster than CD133− tumors, both being GFAP positive (75). The cells of origin of primary glioblastoma may be NSCs, as already stated, whereas secondary glioblastoma must have different cells of origin (102).
CD133 is a marker for embryonic NSCs, for an intermediate radial glia/ependymal cell type and for cells in the early postnatal stage, but not for neurogenic astrocytes in the adult SVZ. In this way, BTSCs have three possibilities of origin: from embryonic CD133-expressing cells, or from CD133+ ependymal cells in the adult, or from CD133− neurogenic astrocytes (103). Since SVZ and SGZ contain astrocytes with stem cell-like features as tumors do, it has been concluded that BTSCs origin from NSCs, and both from the SVZ (104).
Recently, bone morphogenetic protein 4 (BMP4) (105) has been demonstrated to favor astroglial fate in stem cells of the adult (106) and, together with its receptor, to be particularly expressed in CD133+ cells and to show cytostatic properties, reducing stem cell-like cells initiating glioblastoma by 70% and preventing expansion in culture of cells isolated from primary tumor (107). In mice, BMP4 blocks in vivo growth of intracerebral grafting of human glioblastoma multiforme cells. The action mechanism would reside in the triggering of the mothers against DPP homolog 1 (Drosophila) (SMAD) signaling cascade, with translocation to the nucleus and regulation of the target gene transcription (108). These observations, together with the radioresistance of CD133+ cells, may have therapeutic relevance (109).
The Demonstration of BTSCs In Vivo
NSCs are known to reside in perivascular niches, where they are regulated as for self-renewal and fate. Malignant gliomas contain aberrant vascular niches that mimic the normal NSC niche (110). mRNA of a series of stem cell markers was evaluated by quantitative reverse transcriptase-polymerase chain reaction (RT-PCR) in 72 gliomas of different grades, based on the assumption that BTSCs express a series of these antigens (111). It was shown that mRNA of CD133, nestin, SOX2 and musashi-1 correlated with malignancy grades of astrocytic tumors. Immunohistochemically, the four antigens co-stained with GFAP and among themselves, and CD133 and nestin were much more frequently positive than were SOX2 and musashi-1. The four antigens were not associated with Ki-67/MIB.1 labelling index (111). CD133+ cells reached 15.6% in grade IV tumors and this value was similar to the 6-29% found by others (74). It is likely that not all CD133+ cells are BTSCs, even though they may be undifferentiated cells or poorly differentiated cells, not necessarily having the ability for self-renewal and tumorigenesis (111). Nestin increases with malignancy grade (see below) and so does SOX2. Musashi-1-positive cells correlated with malignancy (111); on the other hand, they were also be related to the proliferating activity of gliomas (112). Very importantly, it was observed that musashi-1 mRNA appeared in spheres derived from negative tumors, but it was also lost (111). In our experience, SOX2 and the RE1-silencing transcription factor (REST) stain diffusely positive in paraffin-embedded (Figure 4A, B) and frozen sections of glioblastoma in comparison with nonglial tumors, and also in glioblastoma multiforme-derived neurospheres (Figure 3F). Musashi-1 and CD133 appear to stain positively in frozen sections only and in limited tumor areas or in isolated cells (unpublished data). CD133+ cells were also present in variable percentages among gliomas and in relation to the grade of malignancy. They represented a quota of nestin-positive cells and co-localized with EGFRvIII, correlating with time to progression and overall survival and appearing thus to be a prognostic factor predicting recurrence (83).
Recently, CD133+ cells were demonstrated in formalin-fixed and paraffin-embedded material using an amplifier. They were found in niches, perivascular niches and in single cells and blood vessels in 96% of gliomas, with no correlation with tumor grade or with survival of patients: they were found in nestin-positive as well as in nestin-negative areas, for example in pseudo-palisades and also in relation to small vessels, in all three tumor grades, with no correlation with outcome (82). In glioblastomas, the perivascular location of CD133+ cells forming niches was more evident than in other tumor types. The authors believed that CD133 is not a specific marker of BTSCs, as already demonstrated (75, 76, 113), and it is not predictive of survival.
The antigens studied may not be specific for BTSCs, but they do indicate cell stemness, a stem-cell-like status, or that stemness can be reacquired by dedifferentiated tumor cells. Especially important is the demonstration that nestin+/ CD133+ tumor cells are associated with capillaries in tumor area of high microvessel density, representing perivascular niches for BTSCs. As a matter of fact, CD133+ cells in culture interact with endothelial cells which are known for maintaining self-renewal and a lack of differentiation of BTSCs (114). The association of CD133+ cells and angiogenesis is confirmed in vitro by the high production of vascular endothelial growth factor (VEGF) by glioblastoma cells.
Comparing BTSCs from glioblastomas with those of its recurrence after transplantation in athymic nude mice, it is of considerable significance that the latter develop much more diffuse and aggressive tumors (115).
The Role of Nestin
Nestin expression is typical of neuroepithelial/progenitor cells in rats and humans and it precedes that of vimentin and GFAP during cytogenesis, disappearing from the central nervous sysyem in adults, with the exception of ependymal cells. It can be demonstrated in the progenitor cells during development and in the subependymal layer of adults (116, 117), but with no definite temporal relationship with vimentin and GFAP that appear later during development. Nestin expression is considered characteristic of progenitor cells; however, since it characterizes reactive astrocytes in brain injury as well (98, 118, 119), either these are to be regarded as dedifferentiating in their cell reaction or nestin cannot be taken as a specific indicator of a lack of differentiation (4). It could be indicative of a stem cell-like status without any possibility to distinguish either between NSCs and BTSCs, or between an embryonic or a dedifferentiated phenotype. Interestingly, in culture notch activates nestin promoter, whereas in the SVZ, nestin is produced by the combination of notch and K-Ras (120). In reactive astrocytes, the expression of nestin would represent the embryonic regression of cytoskeleton connected with their morphological plasticity. In hippocampus, it decreases with age (121). There are alternative very complicated interpretations (122, 123).
Generally, nestin and vimentin are not markers of stem cells (74). In pediatric brain tumors, nestin has been found to be expressed in primitive neurepithelial tumors (PNETs), anaplastic astrocytomas and ependymomas, but practically not in low-grade astrocytomas in one series (98) and not expressed in another series of gliomas (98). Using two different antibodies (98, 124), after antigen retrieval, nestin was demonstrated in ependymal cells, in cells of the germinal matrix and in radial fibres of a human fetus, coexpressed with vimentin, but not with GFAP, and in tumor cells of pediatric ependymomas, PNETs, glioblastomas and pilocytic astrocytomas, associated in the latter with vimentin, GFAP and S-100. In our experience (125), nestin is expressed in ependymomas, pilocytic astrocytomas and much more in glioblastomas than in diffuse astrocytomas. It is not expressed in oligodendrogliomas, with the exception of glial fibrillary oligodendrocytes. Endothelial cells were also found to be positive (126) and nestin has been considered to be a marker of rapidly growing endothelial cells (125, 127). It is clearly expressed in microvascular proliferations of glioblastomas (128).
Vimentin was shown to be expressed in all three cell lines from a malignant astrocytoma, whereas nestin was variably positive in the most motile and invasive cells (129). In U373 MG cells, TGFα was observed to reduce GFAP mRNA, but it did not modify that for vimentin and increased that for nestin (59).
Nestin would appear to mark either malignant or less differentiated cells. In glioblastoma, it is variably expressed, and in a complementary way with vimentin and GFAP (125). Proliferating areas containing small cells with hyperchromatic nuclei, invasion areas and recurrences express much more nestin than GFAP (Figure 4C, D, E). In single cells, the complementarity of nestin and GFAP is easily demonstrable and cells of malignant gliomas seem to be distributed according to a spectrum of nestin intensity (125) from lack of differentiation to full differentiation. It is significant that nestin, musashi-1, CD133 and SOX2 are variably coexpressed with GFAP in gliomas (111). Nestin and CD133 associate and increase in malignant gliomas and together correlate with the outcome of tumors (130), but nestin does not seem prognostic within the group of glioblastomas (131).
The Nature of BTSCs
In the light of the cancer theory of somatic mutations accumulating in proliferating cells, in the adult brain there are also proliferating cells capable of accumulating mutations required for transformation. The adult neural stem and/or progenitor cells may be likely candidates and activation of PDGF or EGF pathways in such cells confers on them tumor-like properties (21). A working hypothesis is that tumors are derived by the transformation of undifferentiated precursor cells either of germinal zones or of neurogenetic zones of adult brain. SVZ is the most likely zone and tumors develop near the ventricles and when they develop far from them, this must be because cells have been migrated (132) (Figure 5A).
Neurospheres can also be formed from SVZ and be recognized by the specific assay (11, 133) and population analysis (12); B cells are the true NSCs of the adult (134). Type C cells may be possible source of BTSCs, inasmuch as they possess EGFR and its ligand can produce tumors. The mechanism by which NSCs give rise to BTSCs is still controversial. It is possible that migrating progenitors from NSCs through asymmetric division give rise to generations of migrating tumor cells, so that NSCs do not reside in the tumor, but supply it with progenitor deriving tumor cells (92). Alternatively, asymmetric division may be lost so that stem cells increase, making the tumor grow (135). As a matter of fact, the expression pattern of BTSCs and NSCs do not coincide: the former show higher propagation in vitro and in vivo, are partially independent from mitogens and show tumorigenicity unrelated to specific markers (136). A mathematical model of probable mutations has been put forward, but it is more theoretical than trustworthy (137). The hypothesis of CSCs is strongly based on their self-renewal, without which the pool of initiating cells would be depleted, and on the clonal evolution of BTSCs, which accompanies tumor progression, and involves additional or epigenetic modifications (138). However, the problem of the identification in vivo of BTSCs has not yet been solved. The only possibility is to demonstrate clonogenicity and tumorigenicity of tumor foci drawn from freshly removed tumor samples or to rely on the already discussed demonstration of stemness antigens. Whether BTSCs or cells with a stem cell-like status can be identified in this way cannot be said. Different types of BTSCs have been demonstrated within glioblastoma multiforme, located in different regions of the tumor with different tumorigenetic potentials and genetic anomalies, but deriving from common ancestors. The peripheral pool, grown from a peripheral sample, has the same multipotency, clonogenic ability and self-renewal as a central one, but the latter has a higher clonogenic potential, grows faster, and shows a higher tumorigenicity. Cells from the peripheral pool seem to be derived from those of the central pool, accumulating further genetic aberrations. To find common ancestors to both cell types would be a fruitful goal (87).
The problem has been recently reviewed and three possibilities for the origin of BTSCs have been emphasized: they derive from dedifferentiated glia, from restricted, unipotent progenitors, or from multipotent progenitors (28). A retrograde differentiation of mature glial cells is possible through activated oncogenes or other procedures, provided that these are directed not to immature, but uniquely to mature astrocytes. The target cells, however, must be selected as less mature in the astroglial lineage (28). On the other hand, it is known that reactive astrocytes may undergo embryonic regression and re-express nestin (139). Restricted progenitors can be involved in the process of transformation, including NG2 progenitor cell population which expresses Olig2 and can be re-programmed to produce neurons and astrocytes, besides oligodendrocytes (140). Multipotent progenitors, such as those of SGZ and SVZ, have the potential to form gliomas (95, 141) and the candidates are type B cells which respond to EGF and PDGF. All considered, it has been concluded that BTSCs could be better defined on the basis of functional competence rather than on the identity of the cells which underwent transformation (95).
Signal transduction pathways can modulate growth and differentiation of progenitor cells during development and also in the postnatal life. Notch signaling promotes formation and suppresses differentiation of radial glia (142). EGF, FGF and PDGF promote growth of adult progenitors in SVZ or dentate gyrus. Their receptors activate Ras/Raf/MAPK signaling pathway and involve the phosphoinositide 3-kinase (PI3)/Akt pathway. SHH is a mitogen for granule neuron precursor cells. Other signaling pathways deregulate the cell cycle, suppressing, for example, Bm1 or acting on p53, Olig2, p16, CDK4 etc. Another important aspect of the problem is the ‘tumor competence’ of progenitor cells (143). Since gliomas seem to derive from multipotent type B cells of SVZ, specifically from NG2-positive glia cells, tumors should develop near this region, at least initially. Migration can explain why they are found at a distance from the SVZ at diagnosis.
An extensive analysis of the pros and cons for the hypothesis of the existance of BTSCs leads to the conclusion that it is supported by many experiments, but it does not reconcile with ‘committed astrocyte progenitors or their progenies as cell of origin of gliomas’ (28). A link between adult stem cells and CSCs remains to be shown. Recent observations, however, demonstrate that B cells from SVZ of adult rats progress to amplify transit C cells and transform into tumorigenic cell lines after expansion in vitro. These cell lines expressing nestin, musashi-1 and CD133 continue to proliferate and acquire chromosomal aberrations; when transplanted, they give rise to malignant tumors. PDGFRα is responsible for tumor proliferation and its knockdown by siRNA reduces growth (144). These data go beyond those obtained with the demonstration that adult neural stem cells can be propagated in vitro for over a year, maintaining self-renewal, differentiation, growth factor dependence, karyotype, and molecular profiling without forming tumors in vivo (145). The notion of transformation of normal adult stem cells is a warning in the expansion of NSCs for therapeutic applications. Recently, it was observed that when NSCs from the SVZ are induced to differentiate, they give rise to intermediate progenitors transiently exhibiting multiple glioma characteristics, such as aneuploidy, loss of growth-contact inhibition, alterations in cell cycle and growth factor insensitivity. A subset of cells showed an aberrant expression of class III β-tubulin. Usually, such characteristics disappear with maturation; however, they demonstrate that developmentally intermediate progenitors may give rise to tumor (146).
Neurospheres can be formed not only from glioblastomas, but also from grade II astrocytomas, with a faster growth of the former. Both undergo differentiation in an heterogeneous population of cells, but only neurospheres from glioblastoma invade and destroy neurospheres from normal stem cells from adults (73).
The existence of BTSCs is a working hypothesis, but is very important because it is also relevant to therapy of malignant gliomas: BTSCs, beside being related to angiogenesis, are radio- and chemotherapy-resistant (81, 147). The efficacy of temozolomide, carboplatin, paclitaxel and etoposide seems to depend on CD133+ cells. In primary cultures of glioblastoma, it was shown that mRNA of neural precursors, such as CD90, CD44, nestin etc. was increased in CD133+ cells in comparison with CD133− cells and the former expressed higher levels of breast cancer resistance protein 1 (BCRP1) and methylguanine-DNA methyltransferase (MGMT), as well as of genes inhibiting apoptosis (147). A gene expression profile study demonstrated that the expression signature of drug-resistant glioblastomas was dominated by homeobox (HOX) genes, comprising CD133, as predictors of poor survival, together with EGFR expression (148). In another study, an attempt was made to evaluate cell resistance with CD133+ cells, multidrug resistance 1 (MDR1) and B-cell lymphoma gene 2 (BCL2) (149).
Countering the derivation of BTSCs from normal NSCs (47, 53, 140), a hypothesis holds that they derive from tumor cells through an extreme dedifferentiation process called anaplasia. FISH studies on CD133+ xenografts from a glioblastoma multiforme demonstrated that tumor cells show amplification of EGFR gene, so that both CD133+ and CD133− cells bear the same cytogenetic alterations and therefore they are clonally derived (74). This observation could be in line with the above mentioned hypothesis. However, the existence in the adult brain of neurogenetic sites, such as SGZ and SVZ, may support the hypothesis that tumors arise from mutated developmentally arrested progenitor cells (28). The problem whether stem cells or early progenitors are transformed and show variable differentiation of their progeny during tumor development, or that more differentiated glial cells are transformed by genetic events that lead to a loss of differentiation maintenance, is still debated (150).
Another hypothesis is based on there being a primary originating population composed of progenitor cells (146). Since NSCs of the SVZ infrequently divide, whereas intermediate progenitors have more rapid cell cycle time and tumors are known to originate from a small number of independent mutations (93), it is likely that NSCs accumulate mutations during life which are found in the transformed progeny. Some traits, such as insensitivity to mitogenic factors, aneuploidy, dysregulation of the cell cycle, multilineage capacity and expression of class III β-tubulin, are shared by the tumor-originating population. Progenitors frequently show glioma traits, so that they resemble potential originating cells more than other cells in the development. Using an in vitro model starting from the SVZ of mice, it was shown that intermediate progenitors transiently exhibit multiple glioma characteristics, such as those mentioned above, during normal differentiation. These cells display, therefore, a high capacity for tumor formation (146).
The discussion on the relationship between BTSCs and NSCs is endless and there are observations which cast doubt on there being a direct relation. A series of experiments has been presented in this regard, showing that there are CD133− BTSCs (75) and that endothelial cells are CD133+, so that tumor-initiating cells in xenograft models may be underestimated (151), with the recruitment of non-malignant precursors in the tumor. An interesting observation is that primary glioblastomas contain CD133+ BTSCs and form neurospheres, but CD133− cells and secondary glioblastomas do not seem to contain BTSCs.
To assess the relationship between BTSCs and cell lines is very important. It has been demonstrated that the cells cultured in NBE conditions differ from those grown in serum for morphology, modalities of growth and for potential for multilineage differentiation, clonogenicity and features of NSCs that are typical of neurospheres (100). Another difference is the positive staining for nestin, SOX2 and for stage-specific embryonic antigen-1 (SSEA-1), which is the antigen expressed at the morula stage in mouse embryos (152) and for consistent telomerase activity. NBE cells were clonogenetic and tumorigenetic. Their gene expression profile was similar to that of primary tumors and xenografts and different from that of serum cells; the same is also true for single nucleotide polymorphism analysis, spectral karyotyping and Giemsa banding analysis. On this basis, NBE cells are a more reliable model for understaning biology of glioblastoma multiforme. On the other hand, it was already observed that in NBE cells the gene profile study shows the prevalence of developmental genes in comparison with extracellular matrix genes of serum cells, in addition to a reduced tumorigenicity and invasion in vivo (100, 153).
Molecular Regulation of BTSCs
C-Myc, beside its role in cell proliferation, is an important regulator of stem cell biology, serving as a link between malignancy and ‘stemness’ (154). It is highly expressed in BTSCs, of which it modulates the cell cycle regulators. Knocking down c-Myc by lentivirally transduced short hairpin RNA (shRNA), proliferation of BTSCs is reduced and the cell cycle arrested. With decreased c-Myc expression, neurospheres do not form in vitro nor in tumors after xenotransplantation (155). C-Myc protein was increased in a malignant glioma resembling human glioblastoma, produced in the mouse brain after inactivation of p53 and PTEN by gene targeting and chimera formation methods (86).
EGF is indispensable for the maintenance of BTSCs. In three cell lines, self-renewal and the ability to proliferate were retained only in the presence of EGF and the inhibition of the receptor stopped proliferation and induced apoptosis (156). Other regulators are notch receptors and their ligands on adult NSCs (30, 157), polycomb ring finger oncogene (BMI1), SHH for medulloblastoma, PTEN, Wnt-β-catenin etc. (95).
Cytogenesis and Angiogenesis: The Vascular Niche and the Importance of Hypoxia
The nervous and vascular system share the same origin from the neural tube, since the vascular system originates from the neural crest. The proliferating neuroepithelium is invaded by vascular sprouts from the vascular plexus (158) and, successively, during development endothelial cells and NSCs share the same location (159) and blood vessels are associated with the basal lamina of the SVZ where new neurons are generated (Figure 2D).
The concept of ‘niche’ has been developed recently and refers to a place where stem cells interact with each other and with the extracellular matrix and where a series of factors, such as SHH, Wingless-type protein (Wnt), BMP, FGF and notch regulate proliferation, number and fate of the cells (160). SHH through activation of glioma-associated oncogene (Gli) facilitates survival and proliferation of NSC progeny (161). As a matter of fact, SHH-Gli is activated in gliomas (162, 163), not to mention EGFR, FGF and PDGF. In the niche, the asymmetrical division of stem cells is controlled and deregulated proliferation can start. An important observation is that hypoxia can promote survival and proliferation of NSCs or their progenitors, activating the transcription factor hypoxia inducing factor (HIF-1), with its consequences. In the SGZ and in the SVZ, NSCs are concentrated around blood vessels in areas where progenitor cells cluster and coexist, with pro-angiogenic cells forming the stem cell vascular niche complex (16).
The formation of vascular niches in tumors (Figure 4F) was observed to start with extravasation of circulating bone marrow-derived cells which is the initial step in the sprouting of new vessels within and around tumors (164). These are endothelial progenitor of hemangioblastic derivation, recruited in adult life from bone marrow via circulating tumor-derived VEGF (165). Endothelial progenitor cells differentiate into endothelial cells and are incorporated into the neovasculature (166), recruited by PDGF of endothelial cells to support the outer surface of vessels as pericytes (167, 168). VEGF recruits and mobilizes mesenchymal stem cells (169) that are incorporated into the vessel walls of the tumor before the tumor expands (170), by the aid of BTSCs. Together they form a vascular niche.
Nestin+/CD133+ tumor cells associated with capillaries in areas of high microvessel density represent perivascular niches for BTSCs. In culture, CD133+ cells interact with endothelial cells, which maintain self-renewal and a lack of differentiation of BTSCs (114). There is a cross-talk between endothelial cells and BTSCs as the former stimulate the latter to increase their growth and cell renewal, whereas the latter stimulate endothelial cells to produce VEGF (165). Diffuse staining for CD133 and positively stained isolated tumor cells and endothelial cells have been found in astrocytic gliomas with a different distribution in comparison with that for nestin. CD133 was also distributed around capillaries, without being specific for stem cells or prognostic (82).
Stem cell-like glioma cells (SCLGC) were found to secrete elevated levels of VEGF, which was further induced by hypoxia. In an in vitro model of angiogenesis, SCLGC-conditioned medium significantly increased endothelial cell migration and tube formation, compared with non-SCLGC tumor cell-conditioned medium. The proangiogenic effects of glioma SCLGCs on endothelial cells were specifically abolished by the VEGF neutralizing antibody bevacizumab, which is in clinical use for cancer therapy. Furthermore, bevacizumab displayed potent antiangiogenic efficacy in vivo and suppressed growth of xenografts derived from SCLGCs, but had limited efficacy against xenografts derived from a matched non-SCLGC population (81). Gliomas may arise from transformed NSCs of SVZ, where astrocytes function as NSCs (140) which would play in this way a pro-tumor role. There are observations that migration of NSCs toward tumors has anti-tumor effects (171).
In the origin of gliomas, the role of hypoxia in relation to the vascular niche has been stressed. In a neo-niche surrounding gliomas, recruitment and deregulation of different stem cells for the gliogenetic process could be a key factor; recruitment may play a role in gliomagenesis (160). The hypothesis that recruitment of cells from the adjacent brain and possibly other sites may contribute to glioma formation is not new. It has been based on the paracrine effects of PDGF in animal models of gliomagenesis, on continued adult neurogenesis capable of increasing in response to brain injury, and on the growth factor-rich environment of brain tumors (172).
In primary glioblastomas and in their derived cell lines, cellular and molecular markers occur which are associated with mesenchymal stem cells. This could mean that primary glioblastomas derive from transformed stem cells with properties of mesenchymal stem cells, or that a series of genes controlling mesenchymal properties are activated in glioblastomas (173).
Migration of NSCs toward Gliomas and New Therapies
Targeting of tumor stem cells for therapy is a new goal today, but it has also been found that normal NSCs can target tumors (174). NSCs exhibit tumor-homing capability: immortalized murine NSCs, implanted into glioma-bearing rodents, distributed within and around tumors, even migrating to the contralateral hemisphere (175). Genetically engineered NSCs with their tropism for gliomas may have an adverse effect on the latter (176-179), especially if they are also transduced with herpes simplex virus-thymidine kinase (HSVtk) gene and followed by the administration of systemic ganciclovir (165, 180, 181). Human NSCs implanted in rat brains containing a C6 glioma migrated in the direction of the expanding tumor (182). The same properties are shown by mesenchymal stem cells, injected either into carotid arteries or intracerebrally (183, 184), and by hematopoietic progenitor cells (185).
Endogeneous progenitor cells migrated from the SVZ toward a murine experimental glioblastoma (171). The migrated nestin-positive cells were positive for Ki-67/MIB.1 and 35% of them for musashi-1 (186). Chemokines, angiogenic cytokines and glioma-produced extracellular matrix can play a role in the NSC tropism (187). It is possible to take advantage of the natural capacity of chemokines to initiate migratory responses, and to use this ability to enhance tumor-inhibitory neural progenitor cells to target an intracranially growing glioma (188).
The therapeutic possibilities offered by NSCs are continuously increasing. For example, they can be engineered as sources of secreted therapeutics, exploiting their mobility toward nervous system lesions. They could function as minipumps (189). Rat embryonic progenitor cells transplanted at a distance from a glioma grown in the striatum migrate and co-localize with it. They modify their phenotype, becoming vimentin-positive and reduce the volume of the tumor, demonstrating that a cross-talk exists between them and the tumors (190). It has been shown that hypoxia is a key factor in determining NSC tropism to glioma and that this is mediated by stromal-derived factor 1 and its receptor (SDF-1/CXCR4), urokinase-type plasminogen activator and its receptor (uPA/uPAR), VEGF/VEGFR2 (191). Is it possible to enhance motility of adult NSCs towards central nervous system injury or disease? It seems that EGFR can play a role in this direction with a dual activity, because of its participation in malignant transformation (192). The only limitation to the possibility of migration of neural precursors from SVZ to an induced cortical glioblastoma in mice is age and proliferation of SVZ. Adult mice supply fewer cells than younger mice and this depends on the expression of D-type cyclins. Cyclin D1 is lost during aging and only cyclin D2 remains (193).
Recently, novel treatment strategies using NSCs have been proposed, for example the suicide gene therapy using converting enzyme (194) and others, but new ones will emerge from the studies of NSCs and BTSCs (195). Attention must be dedicated to the possibilities that tumors grow from transplanted NSCs (196).
The Growth of Gliomas and Importance of the Brain Adjacent to Tumor
The currently accepted hypothesis, based on in vitro and in vivo experiments, is that stem cells, progenitors, astrocytes and radial glia are possible sources of tumors. In the stem cell theory the astroglial lineage, from radial glia to astrocytes, as the source of stem cells both in embryos and in adulthood has a special importance (18). SVZ astrocytes have been considered in vivo primary precursors acting as stem cells in vitro (134). These cells may undergo transformation. Experiments in explant cultures showed that human gliomas contain heterogeneous cell populations ranging from precursor cells, expressing for example A2B5 and PSA-NCAM, to mature cells. The former have proliferation and migration capacity, which differ according to the tumor type: very high in glioblastomas and low in pilocytic astrocytomas. These cells are neoplastic glial precursor cells (197).
There is also evidence in the adult central nervous system, with the mechanism of the ‘lineage tracing’ of a transdifferentiation from one cell type into the other and that mutations occurring in gliomas may influence the differentiation or the transdifferentiation state. Cultured p16 and p14−/− astrocytes maintain a diploid status, but shift to a rapidly proliferating status, losing expression of GFAP and acquiring that of nestin (4, 62). This means that p16 and p14 maintain astrocytes in a differentiated status, but in glioblastomas they are inactivated (4) bringing us back to the distinction between the undifferentiated or the transformed phenotype of tumor stem cells. Another important demonstration of how differentiation can be influenced is that gliomas may originate because of the impossibility of progenitor cells to differentiate, as occurs with PDGF on precursors expressing GFAP which assume oligodendroglial morphologies (63).
Studies are being dedicated to the identification of BTSCs and to their recognition in the various glioma grades, also for therapeutic purposes in the frame of the origin of gliomas from NSCs. Protein and mRNA expression of a series of stem cell markers were studied in a number of gliomas and compared with those of normal nervous tissue. CD133, nestin, SOX2, musashi-1, CXCR4, Flt-4/VEGFR-3 and CD105/endoglin were found to be highly variable and increased in gliomas in comparison with normal brain, but only nestin, CD133, musashi-1 and SOX2 in a grade-dependent manner. They were co-stained in complex patterns with GFAP and less frequently with Ki-67/MIB.1 (111). Gliomas therefore contain a considerable proportion of cells expressing stem cell markers, but it remains to be definitely clarified the relationship between BTSCs and NSCs which are attracted to the tumor (111, 127).
There is no evidence that tumors can develop from the proliferating reactive glia; however, they might originate from radial glia, into which differentiated astrocytes can regress under certain stimuli (139) and radial glia can proliferate. Bone marrow stem cells must also be considered as possible source of tumors, because of their capacity for differentiating along the neuroectodermal line (198). Interleukin-8 (IL-8), transforming growth factor (TGF)-ss1 and neurotrophin-3 (NT-3) could contribute to the glioma-directed tropism of human marrow stromal cells. Since VEGF is another marrow stromal cell-attracting factor secreted by glioma cells, there could be support to the hypothesis that gliomas use their angiogenic pathways to recruit mesenchymal progenitor cells (169). The hypothesis is that CXCR4+ tissue-committed stem cells of the bone marrow, circulating in the peripheral blood (199), could participate to glioma tumorigenesis (163).
The fundamental question is whether gliomas initiate in situ from abnormal niches or are composed of transformed stem cells migrating from the classic niches with which they subsequently lose continuity with the development of the tumor. The discussion must take into account the suggested possibility that stemness could be a transient and reversible status with multipotentiality, which would confer plasticity to stem cells rather than self-renewal (200). On this basis, the hypothesis has been put forward that any site in the brain may contain ‘plastic’ cells because of hypoxia. In these sites, cells may acquire stem cell status through hypoxia and its cascade with HIF-1, VEGFR, EGF, PDGF etc. Stem cells may escape the original niche, attracted by a hypoxia-activated neo-niche which warrants proliferation of transformed stem cells. These participate in the gliomagenesis. The neo-niche attracts surrounding differentiated cells with plastic features, committed to becoming astrocytes or oligodendrocytes. CXCR4+ neural-committed stem cells from bone marrow may be recruited as well (160). The possibility that dedifferentiated glioma cells may acquire stemness and express its antigens has already been depicted.
An area of the greatest interest not only for the biology and pathology of gliomas, but also for neurosurgery, radio- and chemotherapy because of its controversial imaging is that generally indicated as ‘brain adjacent to tumor’ (BAT), of 1-3 cm around tumors. Besides the intense movement of motile and invasive tumor cells, the BAT is an area of great cell traffic where many factors and regulators play a role (Figure 5B). An example is given by the most recently described Na(+)/H(+) exchanger regulatory factor 1 (NHERF-1), which sustains glioma cell migration and invasion (201). Moreover, glutamate accumulates in the BAT because of cystine/glutamate exchange, with consequences on tumor growth and epileptic seizures (202). Tumor cells swarm from the tumor in every direction, and local astrocytes respond becoming reactive astrocytes, showing embryonic regression and expressing nestin. It cannot be excluded they may transform into tumor cells. Local astrocytes may undergo the same transformation as they belong to the astrocytic lineage susceptible to acting as NSCs that can enrich the tumor by neoplastic conversion. The BAT is usually rich in activated microglia cells and macrophages, sources of tumor necrosis factor (TNF)-1 family members and chemokines and is the place where secondary niches for NSCs may arise, where hypoxia develops and acts through HIF-1. NSCs and transformed NSCs may be attracted by the tumor and migrate towards it, not to mention stromal bone marrow stem cells that reach the tumor in the brain via blood stream for angiogenesis. In conclusion, the BAT is a crucial area for tumor growth and diffusion and for its delimitation by neuroimaging conventionally used for radiation therapy. It is a passageway to and from the tumor of cells with different potentialities and genetic regulations, subjected to the surveillance of the blood-brain barrier, with its cantless molecular protective mechanisms and dysfunctions, and of the immune system. This has an enormous importance for diagnostics by neuroimaging, aiming at the definition of tumor borders, and for therapeutic strategies focused on the local control of the tumor.
Conclusion
The study of BTSCs, provided that they exist as a cell type and their definite demonstration will be given, has been of great help in approaching the problems concerning the origin of gliomas, their formal genesis, growth and diffusion and therapies. The doubts on their existence as a cell type did not dispel the notion that a functional status of stemness can be reached in different ways. Often the term “stem cell-like” is used for indicating a cell population behaving like real stem cells and responsible for growth, recurrence and resistance to radio- and chemotherapy. Among the different modalities, that of Alvarez-Buylla et al. opened the door to a wide possibility of origin of NSCs and therefore of BTSCs, including adult cells belonging to the astrocytic lineage; it does not exclude the possibility of the origin of gliomas from adult astrocytes. Practically, the interpretations of BTSCs as a status reached by embryonic regression of dedifferentiated tumor cells would respect all their perogatives without the necessity of there being a special cell type.
One of the most important consequences is the possibility that the BTSC concept offers to overcome cell resistance to therapies. Many attempts are continuously being made in order to achieve positive results, either through the molecular manipulation of cells or their direct therapeutic attack.
Among the different possibilities for growth and diffusion of glioma, BTSCs and NSCs represent a crucial parameter in their diagnosis and treatment.
Acknowledgements
These studies were supported by a grant from Compagnia di San Paolo, Turin, Italy.
Footnotes
-
↵* Presented at The 8th International Conference of Anticancer Research, Kos, 17-22 October 2008.
- Received August 13, 2009.
- Revision received April 22, 2010.
- Accepted April 28, 2010.
- Copyright© 2010 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
References
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵