Background
The cancer relapse and mortality rate suggests that current therapies do not eradicate all malignant cells. In this sense, there is increasing evidence that many types of cancer contain their own stem cells: cancer stem cells (CSCs), which are characterized by their self-renewing capacity and differentiation ability [
1]. The study of haematological disorders shed light on the relationship between cancer and stem cell compartments, and the mechanisms by which CSCs might appear and change during the progression of the disease [
2,
3]. However, the evidence for the existence of CSCs in solid tumours has been more difficult to find because of the lack of specific cell surface markers. During the last years, different cancer cell subpopulations from selected types of human solid cancers have been identified (breast [
4], brain [
5‐
7], colon or colo-rectal [
8‐
10], head and neck [
11] and pancreatic cancer [
12]). These authors, through the use of cell culture, FACS and/or MACS methods, have been able to identify different cell populations within the tumour showing hallmarks of stem cells. This stem cell potential, including self-renewal and lineage capacity, was demonstrated by serial transplantation experiments in animal models. Specifically, the investigation of solid tumour stem cells has gained momentum particularly in the area of gliomas, the most common type of brain tumours. In this group, glioblastoma multiforme is the highest-grade glioma [GBM; grade IV] and is manifested by morphological, genetic and phenotypic heterogeneity [
13‐
15]. Two major aspects of glioma biology that contributes to its awful prognosis are the formation of new blood vessels through the process of angiogenesis and the invasion of glioma cells, the hallmarks of GBM [
16]. In addition, these abnormal blood vessels have also been shown to create a vascular niche that houses glioma stem cells [
17].
Despite of the recent advances, two-year survival for GBM patients with the most favourable treatment is less than 30%. Even in those patients with low-grade gliomas therapy is almost never curative. Recent studies have confirmed the existence of a small portion of glioma cells with characteristics of neural stem cells [
1]. In general, this fraction is characterized by its neurosphere-forming ability, its strikingly increased drug resistance, and finally, by its ability to express surface markers that are oftenly used for their FACS-based isolation [
5,
6]. With the implantation during this last decade of the NS forming assay as a robust method for the isolation of neural stem cells [
18], it has become widely accepted that adult mammalian brain harbours a pool of NSCs responsible for the persistent neurogenesis, seen in limited adult brain regions, such as the sub-ventricular zone, olfactory bulb and hippocampal dentate gyrus [
19].
However, it should be borne in mind that the NS assay is not the most suited source of primary adult stem cells for transcriptomic analysis since cells are selected based on its in vitro proliferation capacity in the presence of cytokines and growth factors in their cell cultures such as EGF and FGF.
At the end of twenty century, two independent laboratories could identify and isolate human central nervous system stem cells using antibodies against CD133 [
20,
21]. This protein, named prominin, identifies a subset of human foetal brain cells distinct to human haematopoietic stem cells, which are also CD133+ but are also CD34-bright [
22]. This subset of human CD133+ fetal brain cells is capable of neurosphere initiation, self-renewal, and multilineage differentiation at the single-cell level [
20]. The CD133+ cells can differentiate
in vitro to neurons and glial cells, and their transplantation into the lateral ventricles of newborn NOD-SCID mouse brains resulted in specific engraftment in numerous sites of the brain [
20,
21,
23].
The CD133 marker is a five-transmembrane protein which is expressed in different type of progenitors as human fetal brain cells or human hematopoietic stem cells [
20‐
22]. In brain tumours the proportion of these CD133+ cells represent a minority of the tumour cell population and are also capable to initiate tumour formation
in vivo. Although it have also been reported that a proportion of these tumours could be maintained by CD133- cells [
24], there are several evidences showing how this small fraction of CSC which forming NS, can also be isolated using CD133 as a selection marker [
6].
In the present study, we have analysed thoroughly the molecular signature of eight fresh primary GBMs focused on its CD133 positive and negative cells. Importantly, all tumours were studied before any treatment of the patient and without previous tumour cell culturing. In addition to the expression analysis of the FACS-sorted cells, we have also performed genome-wide analysis by CGH-arrays, FISH studies at PTEN and EGFR loci, and MLPA at the MGMT promoter. The results obtained concluded that the gene expression signature of CD133+ discriminate common genes to all samples involved in two main characteristic pathways deregulated in GBMs, angiogenesis and invasiveness. However, CD133+ gene expression profile also allowed distinguishing between two different GBM subtypes in higher or lowering proliferative tumours. The molecular biology and the expression signature of these CD133+ cells that drive and support the tumour growth will shed light on the development of fresh and specific treatment strategies.
Discussion
Glioblastomas are the higher-grade gliomas with fast progression and unfortunate prognosis. Recent studies have demonstrated in these tumours the existence of a small fraction of cancer cells endowed with features of primitive neural progenitor. Although some observations pointed out towards the involvement of CD133- cells in tumour maintenance [
24], several studies have involved the CD133
+ cells as the brain tumour initiating cell [
6,
41,
42]. In any case, studies performed in order to characterize the glioblastoma stem cell have been carried out using
in vitro, cultured tumour cells. While these cultured cells present the capacity to form NS, essential pathways in cell/tumour biology could likely be altered as a direct consequence of the cell culturing such as cell-cell adhesion, cell-niche junctions, exposure to mitogen activation, rapid division of the cells etc.
To gain insight into the characterization of these cells, we examined directly for the first time CD133+ cells by FACS-based purification from ex-vivo primary tumours without the intervention of cell culturing or any prior expansion procedure.
Despite that the cohort of tumours analysed was not very large, we could find a correlation among clinical history, response to treatment and genomic alterations in two samples (G4 and G11). Both of these tumours showed the highest content of CD133+ cells, the lack of response to treatment and similar chromosomal alterations (multidimensional scaling analysis reflected relationship among these parameters). However, this correlation was not further supported by transcription profiling of CD133+ and CD133- cells. Indeed, non-supervised analysis of CD133+ vs. CD133- gene expression (Figure
4) showed that only G9 and G11 samples were grouped together and apart from the rest, being the tumour location the only biological feature able to distinguish them (see Table
1). Despite of this low number of samples, multidimensional scaling analysis established again a relationship between G9/G11 tumour location and their gene expression.
To understand the biological properties of the CD133+ compartment, we sought to identify common gene signature by the comparison of CD133+ vs. CD133- cell populations. This array-based analysis led us to the identification of gene profiles with common up-regulated and down-regulated genes. Up-regulated genes such as
COL1A1,
COL1A2,
PGF [
38] or
TGFB1 [
43], suggested an important role of these compartment in blood vessel formation, angiogenesis, permeability and invasiveness, essential functions in tumour progression [
38‐
40]. Significantly, most of these up-regulated genes encoded secreted proteins involved in autocrine and paracrine signalling, like
TGFBI, a pleiotropic cytokine that, among other functions, can induce the dissociation of VE-cadherin junctions between endothelial cells which could favour mature tumour or GBM cells migration [
43]. Up-regulation of these genes in putative CD133+ stem cells would help to increase the mobility of cancer stem cells through the brain, which is consistent with the high invasive characteristics of these tumours and their high possibility to colonize the adjacent area. It is also worthy to mention in this same regard, the importance of the microenvironment in the stem cell/cancer stem cell maintenance, as has recently been pointed out with the identification of the perivascular niche in grade I-IV astrocytomas [
44]. Several evidences suggest that normal neural stem cells, and likely also neural cancer stem cells, exist within protective niches as the vascular niches, into which endothelial cells secrete factors that regulate neural stem cell function [
45,
46]. This raises the question of whether CSCs could be located and regulated by these microenvironments. Calabrese et al. proposed that the tumour microvasculature generates specific niche microenvironments promoting the maintenance of CSCs [
47]. Recent studies using orthotopic glioblastoma xenografts suggest that CSCs secrete proangiogenic factors that promote the recruitment and formation of tumour blood vessels [
48] that significantly facilitates brain tumour growth and invasion. Our gene expression findings in
ex-vivo CD133+ isolated cells clearly support this result.
High expression in the CD133+ compartment of genes such as
LRRFIP1, transcriptional repressor of
EGFR [
49], would support the idea of
EGFR gene as a secondary event in the process of GBM development by promoting infiltration and mediating resistance to therapy. In this same scenario, the positive regulation of the tumour suppressor gene
TMEFF2 [
50] in the potential CD133+ stem cell compartment in these GBMs, could also operate as a late event in the initiation of neoplastic progression. In fact, low levels of
TMEFF2 and other genes responsible for tissue or cell assembly in the CD133- compartment would promote the down-regulation of cell to cell interactions and junctions, providing a molecular mechanism for the highly invasive nature of the GBM.
The second group of genes, commonly down-regulated in all CD133+ vs. CD133- cell from human
ex-vivo GBM samples (that means, over-expressed in the CD133- compartment) were found to be associated to cell assembly, neural cell organization and neurological disorders. That is the case of genes such as
GNB2L1, an anchor protein involved in adhesion and migration of human glioma cells [
51],
DPYSL2, a promoter of microtubule assembly and neuronal development [
52],
TUBA1A [
53] or
CFL, which controls cell migration and cell cycle progression [
54,
55]. This group of genes plays important roles in cell migration, cell polarity and actin polymerization (Figure
6). In this same oncogenic scenario, it would be interesting to mention the deregulated expression of
HIF-1 gene. This gene which is down-regulated in most of the CD133+ samples analysed, is involved in tumour angiogenesis and cell growth [
56], and could play some role in the later events that drive tumour progression. In this regard, recent studies have demonstrated that HIF-1 protein stabilization contribute to tumour angiogenesis, one of the main characteristics of primary GBMs [
16]. Mutations in metabolic enzymes, in particular isocitrate dehydrogenase enzymes (IDH1 and IDH2), have been shown to be involved in glioma development and would facilitate HIF-1 protein stabilization [
57,
58]. The negative deregulation of the
HIF-1 gene that we have observed in most
ex-vivo CD133+ cells in this work, also support this idea.
A notable feature of the gene expression pattern of CD133+ cells was the differential expression of 40 genes that divide GBM samples in two opposite molecular signatures. The classification of these 40 genes according to their function (Figure
8 and Table
4) pointed to their implication in cell growth, cell death, DNA replication, recombination and, definitively, in cell proliferative control. Amongst these genes we wanted to emphasize the differential expression of the gene coding for vimentin (
VIM), an intermediate filament of the mesenchymal lineage involved in migration, cell signalling, cancer and neurological disease [
59,
60]. Some other genes differentially expressed in this pool of cells CD133+ and also involved in cancer and neurological disease are
RPS4X,
RPS3A or
TUBA1B (Table
4). Another relevant member of the top 40 list of genes was
HUWE1, a pleitropic ubiquitin ligase that participates in a wide variety of biological functions related to cell proliferation such as cell growth/death, and DNA replication, and that has been described to be deregulated in different carcinomas [
61] (see Figure
8 and Table
4). Interestingly, this deregulated gene has also been reported to be an important control gene for the proliferation capacity of embryonic NSC in the mice [
62].
GLUL encodes the glutamine synthetase, a metabolic enzyme required for the maintenance of the energy balance and that when mutated causes severe malformations and neonatal death [
63]. Finally,
PLK1, the mitotic kinase par excellence, modulates mitosis entry and promotes cell transformation upon upregulation as an oncogene [
64‐
66]. These differentially regulated molecules must be playing pivotal roles in keeping the tumour cells in a switch-on state that enables them to survive, proliferate and invade the surrounding healthy tissue.
Table 4
Functional classification of 40 differentially expressed genes in CD133+ vs.CD133- GBM samples
Protein Synthesis | 1,14E-10-1,14E-10 | EIF2AK4, RPL22, RPS4X, RPS3A, RPL27A, RPL7A, RPL39, RPL23A, RPL41, RPL7 |
Cancer | 1,17E-04-4,38E-02 | HUWE1, TRA2B, VIM, PLK1, TUBA1B, ACTG1, RPL7, TPT1, RPS4X, RPS3A, H3F3A, RPS16, GLUL, HSP90AA1, CLEC2D |
Cellular Growth and Proliferation | 1,94E-03-2,89E-02 | HUWE1, PLK1, RPL23A |
Cell Death | 1,99E-03-3,52E-02 | HUWE1, RPS3A, HSP90AA1, VIM, PLK1 |
Cell Morphology | 1,99E-03-2,36E-02 | VIM |
Cellular Assembly and Organization | 1,99E-03-3,71E-02 | VIM, PLK1, ACTG1, RPL7 |
DNA Replication, Recombination, and Repair | 1,99E-03-3,71E-02 | HUWE1, VIM, PLK1 |
Cell-To-Cell Signalling and Interaction | 3,97E-03-9,9E-03 | VIM |
Cellular Function and Maintenance | 5,49E-03-2,94E-02 | EIF2AK4, HSP90AA1, VIM, ACTG1 |
Cellular Development | 5,95E-03-1,97E-02 | EIF2AK4, HSP90AA1, VIM |
Neurological Disease | 9,02E-03-4,28E-02 | TPT1, RPS4X, RPS3A, RPL39, VIM, ACTG1, NSMAF, TUBA1B, CALM2 |
Skeletal and Muscular Disorders | 9,02E-03-9,02E-03 | TPT1, RPS4X, RPS3A, VIM, TUBA1B |
Cellular Movement | 1,38E-02-1,38E-02 | VIM |
Gene Expression | 2,36E-02-3,13E-02 | PABPC1, PLK1 |
In brief, the results obtained in this study revealed the presence in CD133+ cells from primary glioblastoma of a common gene expression signature involved principally in the promotion of proangiogenic and invasive programs. Additionally, CD133+ gene expression pattern led us to discriminate between two different GBM subtypes in higher or lower proliferative tumours. The molecular biology and the expression signature of these CD133+ cells that drive and support the tumour growth will shed light on the development of new treatments to fight against GBMs.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JLG participated carrying out CGH, FISH and methylation studies. MPC carried out qPCR assays, analysis and interpretation of array data and has been involved in drafting the manuscript. JAGM provided GBM samples and clinical data from patients. FGR carried out gene expression assays and performed gene expression analysis. JO, OB and MS contributed to anatomy-pathological diagnosis of GBM samples. JMHR and RGS have been involved in revising it critically for important intellectual content. MSM conceived of the study, its design, coordination and has been involved in drafting the manuscript. All authors read and approved the final manuscript.