The presence of a CD133+/nestin+ population in brain tumors suggests that a normal neural stem cell may be the cell of origin for gliomas. We have identified human CD133-positive NSCs from adult glioma tissue and established them as long-term in vitro cultures human neuroglial culture (HNGC)-1. Replicative senescence in HNGC-1 led to a high level of genomic instability and emergence of a spontaneously immortalized clone that developed into cell line HNGC-2 with features of cancer stem cells (CSCs), which include the ability for self-renewal and the capacity to form CD133-positive neurospheres and develop intracranial tumors. The data from our study specify an important role of genomic instability in initiation of transformed state as well as its progression into highly tumorigenic CSCs. The activated forms of Notch and Hes isoforms were expressed in both non-neoplastic neural stem cells and brain tumor stem cells derived from it. Importantly, a significant overexpression of these molecules was found in the brain tumor stem cells. These findings suggest that this model comprised of HNGC-1 and HNGC-2 cells would be a useful system for studying pathways involved in self-renewal of stem cells and their transformation to cancer stem cells.
Disclosure of potential conflicts of interest is found at the end of this article.
The adult brain is composed of various differentiated and specialized cell types like astrocytes, neurons, oligodendrocyte, and ependymal cells. Studies by Reynolds and Weiss in the early 1990s showed that certain areas of the adult brain like hippocampus, subventricular zone (SVZ), and olfactory bulbs contain a small percentage of undifferentiated and multipotent neural stem cells and glial progenitor cells . Recent studies provide compelling evidence that transformation of mitotically active undifferentiated neural precursors within the brain are the most likely targets for transformation evolving into cancer stem cells (CSCs) [2, 3]. The CSCs are multipotent cells capable of forming heterogeneous tumors in immunodeficient mice at a high frequency . Most human tumors have now been shown to contain a subpopulation of cells that display cancer stem cell characteristics like leukemia , breast cancer , melanoma , lung cancer , and brain tumors [9, 10].
Gliomas are the most heterogeneous group of brain tumors comprised of morphologically diverse cells that express a wide variety of neural lineage markers . The characterization of brain tumor stem cells (BTSCs) has been on the basis of expression of neural stem cell markers like CD133 and nestin, ability to form neurospheres, capacity of cells to self-renew, proliferate, and differentiate to recapitulate the original tumor phenotype . Importantly, for the cells to be categorized as cancer stem cells, as few as 100 CD133+ cells should be able to form brain tumors in immunodeficient mice .
The similarities between the self-renewal mechanisms of stem cells and cancer stem cells have led to the recently proposed cancer stem cell model hypothesis, which states that brain tumor stem cells arise from intrinsic mutations in normal stem/progenitor cells leading to their uncontrolled proliferation . We have earlier described a model system useful for understanding gliomagenesis, comprising of a long-term in vitro culture of human neuroglial culture (HNGC)-1 and an established cell line, HNGC-2, derived from the same human adult glioma tissue . The HNGC-1 is an early passage cell culture that is contact-inhibited, nontumorigenic, and noninvasive, whereas HNGC-2 derived from HNGC-1 by spontaneous transformation is a rapidly proliferating, anchorage-independent, highly tumorigenic, and invasive cell line. Interestingly, in this study, we demonstrate that HNGC-1 cells are nontumorigenic stem cells and provide evidence that genomic instability is causal for emergence of spontaneously immortalized cancer stem cell line HNGC-2 from HNGC-1. Our data support the recent concept that glioma stem cells expressing CD133 (Prominin-1), a marker for both neural stem cells and brain cancer stem cells, are responsible for the highly proliferative and malignant nature of glioblastoma (GBM).
Both HNGC-1 and HNGC-2 cells show activation of Notch and of their transcriptional regulators of the Hes family, indicating the involvement of Notch signaling pathway in survival and growth in both nontumorigenic stem cells and brain tumor stem cells. Interestingly, there is a constitutive over-activation of the Notch pathway in HNGC-2 cells, implicating an important role for Notch signaling in brain tumor stem cells.
The frequency of spontaneous immortalization and transformation is extremely rare in human adult stem cells and is reported recently only in mesenchymal stem cells . Our study is the first to depict spontaneous transformation of human nontumorigenic stem cells to highly transformed and tumorigenic cancer stem cells and demonstrate the role of genomic instability in their evolution. The HNGC-1 and HNGC-2 cell system would thus serve as a useful model for defining the spontaneous tumorigenic events leading toward progression to GBM.
Materials and Methods
The cell lines HNGC-1 and HNGC-2, established from a human glioma tumor tissue, have been previously described . The cell lines were maintained in neurobasal medium (NBM) with B27 supplement at 37°C in 5% CO2. The HNGC-1 cells in passages p8–p12 and HNGC-2 cells in passages p30–p34 were used for the study.
Generation of Neurospheres
Neurospheres were generated from both HNGC-1 and HNGC-2 cells by seeding cells in NBM supplemented with basic fibroblast growth factor (bFGF) (20 ng/ml) and epidermal growth factor (EGF) (20 ng/ml) (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) and incubating them for 10 days. Primary neurospheres were mechanically dissociated, and cells were seeded at a density of 1 × 103 cells per milliliter for developing secondary neurospheres, wherein the primary neurospheres were mechanically dissociated into single cells, followed by seeding of the isolated cells in NBM with bFGF (20 ng/ml) and EGF (20 ng/ml) and incubating for 10 days at 37°C in 5% CO2.
The HNGC-1 and HNGC-2 cells were seeded at low density on glass coverslips and grown for 24 hours to achieve semiconfluent cultures. The cells were fixed with methanol for 10 minutes at 4°C, washed with phosphate-buffered saline (PBS), and blocked with 1% bovine serum albumin (BSA) in PBS. The cells were incubated with the following primary antibodies for 2 hours at 4°C: CD133 (1:50; Miltenyi Biotec, Bergisch Gladbach, Germany, http://www.miltenyibiotec.com), nestin, musashi, bmi-1, Sox-2 (1:100; Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com), nucleostemin (1:100; Chemicon, Temecula, CA, http://www.chemicon.com), p53 (1:100; Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com), phospho-p53, p21, Notch-1, Notch-2, Notch-3, Hes1, Hes2, Hes4 (1:100; Chemicon), ataxia-telangiectasia mutated (pATM), 53BP1, γH2AX1, Chk1, Chk2, Brca1, Brca2, Mre11, RAD52, ARK-1 (1:20–1:100; Cell Signaling), and Smc1 (1:100; Calbiochem, San Diego, http://www.emdbiosciences.com). Later, the cells were incubated with corresponding secondary antibodies tagged with Alexa595 (Molecular Probes, Carlsbad, CA, http://probes.invitrogen.com) for 30 minutes at room temperature. Appropriate isotypic antibodies were used as negative controls.
Induction of Neural Differentiation
Differentiation of neurospheres was induced by withdrawing mitogens bFGF or EGF from the NBM and allowing them to differentiate in absence of growth factors for a week. On day 7, the cells were fixed in methanol and stained using antibodies to glial fibrillary acidic protein (GFAP) (astrocyte marker), Tuj-1 (neuronal marker), anti-oligodendrocytes—clone-NS1 (oligodendrocyte marker), and tyrosine hydroxylase (DOPA-positive neuron) (Chemicon) followed by AlexaFluor conjugated secondary antibodies (Molecular Probes).
The HNGC-1 cell line was G-banded using standard procedures. Briefly, the HNGC-1 cells were harvested following incubation with 100 μl of 10 μg/ml colcemid (Gibco, Grand Island, NY, http://www.invitrogen.com) for 30 minutes. Cells were then given hypotonic treatment with 0.075 M KCl at 37°C for 20 minutes, pelleted, and fixed with 1:3 acetone:methanol fixative. GTG-banding was done on 3-day-old slides using standard protocols as described .
Soft Agar Assay
The clonogenic potential of cell lines HNGC-1 and HNGC-2 was assessed by soft agar assay . Each set was plated in triplicate, and the assay was performed at least three times. The colonies were counted using ×20 objective in 10 different fields to acquire the average number of colonies per cell line.
Senescence assay was performed with HNGC-1 and HNGC-2 cell cultures using Senescence β-Galactosidase Staining Kit (Cell Signaling) according to the manufacturer's protocol. Briefly, cells were grown in a 35-mm plate for 48 hours, washed with PBS, fixed, and stained using 1 ml of staining solution in citrate buffer (pH 6.0) overnight at 37°C. Cells were observed for development of blue color under a microscope using ×10 objective .
In Vivo Tumorigenicity and Engraftment Assays with Green Fluorescent Protein Transfected HNGC-2 Cells
The HNGC-1 and HNGC-2 cells were stably transfected with plasmid-enhanced green fluorescent protein (EGFP)-N1 (Clontech, Palo Alto, CA, http://www.clontech.com). G418 resistant clones were selected after 2 weeks and expanded into a stable cell line. The brain engraftment assay was performed with green fluorescent protein (GFP)-transduced HNGC-1 and HNGC-2 cells cultured in NBM medium as well as in cells grown in NBM with EGF and bFGF by injecting cells beginning from 1 × 102 to 1 × 107 in 10 μl of PBS intracranially in nude mice and examined for behavioral changes such as abnormal movement and paralysis. Animal experimentation was done in accordance with the rules and regulations of the Institutional Animal Ethics Committee of NCCS.
Immunohistochemistry of Mouse Brain Tumor
The brain tumor tissue from mice injected with EGFP-N1 transfected HNGC-2 cells was fixed and processed for immunocytochemical staining. The sections were stained with GFP, GFAP, nestin, and CD133 antibody followed by Alexa595/Alexa488 conjugated corresponding secondary antibodies . Also, the sections were stained with hematoxylin and eosin (H&E) for histopathological studies.
Telomeric Repeat Amplification Assay
Telomerase activity was determined using polymerase chain reaction (PCR) based TRAPeze Telomerase Detection Kit (Chemicon) according to the manufacturer's instructions. Total proteins (1.5 μg) extracted from HNGC-1 and HNGC-2 cells were resolved on 10% nondenaturing polyacrylamide gel by electrophoresis and stained with ethidium bromide for observation under ultraviolet light. Telomerase-positive cells generated by us served as a positive control in the assay.
Genomic Instability Using Intersequence Simple Repeat-PCR
Intersequence simple repeat (ISSR)-PCR was performed by using a method similar to the one used for oral squamous cell carcinoma . Genomic DNA was extracted from HNGC-1 and HNGC-2 cells by standard method. We amplified 500 ng of the DNA in a 50-μl reaction containing 1× PCR buffer, 1.5 mM MgCl2, 200 μM deoxyribonucleotide triphosphate, 2.5 μM each of the base-anchored dinucleotide repeat primers ((CA)8RG, (CA)8RY, and (CG)4RY [Microsynth, Balgach, Switzerland, http://www.microsynth.ch]), and 0.5 units of Taq DNA polymerase (PerkinElmer Life and Analytical Sciences, Boston, http://www.perkinelmer.com). The PCR products were analyzed on an 8% nondenaturing polyacrylamide gel with 5% glycerol.
The HNGC-2 adherent cultures in passages between 35 and 40 (1 × 106 cells) and cells grown in NBM medium with growth factors were used for analysis of CD133. Expression of Notch-1 intracellular domain (NICD) using flow cytometry was studied only with HNGC-2 adherent cells. Briefly, the cells were washed with 1× PBS, detached using a cell scraper for adherent culture, and dislodged mechanically for neurospheres. The cells were fixed using 3.7% paraformaldehyde. Later, the cells were blocked with 1% BSA for 15 minutes and stained with Notch-1 NICD antibody and CD133 at 1:100 dilution for 1 hour at 4°C. Rabbit IgG isotype control was used to detect any nonspecific fluorescence. The cells were washed with 1× PBS and incubated with fluorescein isothiocyanate-conjugated goat anti-rabbit antibody for 30 minutes at 4°C. The cells were analyzed for the expression of NICD and CD133 by using FACSVantage and FACSAria and analyzed using Cell Quest and DIVA Software (both from Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com).
HNGC-1 Cells Exhibit Features of Self-Renewal, Express Neural Stem Cell Markers, and Are Multipotent
The functional characterization of NSCs in vitro is based on their ability to proliferate, exhibit self-renewal, generate progeny through transient amplification of a population of progenitor cells, and retain their multilineage potential over time. Initial experiments were aimed to investigate whether the HNGC-1 cells possessed properties similar to those exhibited by stem cells. The HNGC-1 cells appeared as small, flat, nonrefractile cells with extensive self-renewal capacity and could be propagated in culture under serum-free conditions over limited passages (p30–p32) (Fig. 1Aa). The cells plated in the presence of EGF and bFGF formed small clonal clusters referred to as neurospheres within 3–4 days. These neurospheres increased in size and, by day 10, measured around 22 ± 2.64 μm (Fig. 1Ab) and were CD133-positive (Fig. 1Ac). The cells obtained from neurospheres demonstrated the potential to generate new secondary neurospheres that measured around 20 ± 3.93 μm, indicating their capacity for self-renewal. The neurosphere-derived cells were strongly positive to a multitude of neural stem cell markers—CD133, nestin, musashi-1, bmi-1, Sox-2, and nucleostemin (Fig. 1B). Interestingly, nestin, besides showing a strong filamentous expression in the cytoplasm, displayed a strong nuclear staining. The early neural markers Sox-2, bmi-1, and nucleostemin, involved in self-renewal and proliferation of stem cells, had a specific nuclear localization, whereas musashi-1, a neural RNA binding protein considered an important marker for self-renewal of NSCs, showed nucleocytoplasmic staining. The expression of these molecules was also confirmed at the RNA level by reverse transcription-PCR using gene-specific primers (data not shown). The neurosphere-derived cells did not stain positive for any of the neural differentiation markers. The HNGC-1 cells were multipotent as the neurospheres grown in neural differentiation-inducing medium displayed typical morphological differentiation toward all three neural lineages—astrocytic, neuronal, and oligodendrocytic as assessed by positivity for GFAP (astrocytic), Tuj1 (neuronal), and oligodendrocyte antigen, respectively. The cells cultured in dopaminergic medium showed enrichment of tyrosine-hydroxylase-positive neurons (supplemental online Fig. S1). The extent of differentiation determined as percent positivity to the three neural lineages studied across three different passages in HNGC-1 cells was astrocytic (52%), neuronal (40%–45%), and oligodendrocytic (10%) (supplemental online Fig. S2). The differentiated cells, as expected, did not express any stemness-related genes, and less than 1% of cells showed coexpression of any lineage markers. The karyotypic analyses and clonogenicity assays in soft agar indicated that the cells exhibited a normal karyotype, were anchorage-dependent, and lacked transforming activity in vitro (supplemental online data S3a, S3b). Collectively, the expression of stemness- and self-renewal-associated molecules and the potential of HNGC-1 cells for multilineage differentiation provide important evidence that they are nontumorigenic stem cells.
The Spontaneous Transformation of HNGC-1 Stem Cells Led to Development of an Immortalized Clone
During long-term continuous culture of HNGC-1 under serum-free conditions, around passages p28–p30, the cells reached senescence (Fig. 2Aa). We found that most cells underwent cell cycle arrest and showed positive β-galactosidase staining at pH 6 (Fig. 2Ab). Interestingly, during the onset of senescence, the cells underwent a period of crisis followed by emergence of a spontaneously immortalized clone growing over the adherent population. This clone was composed of small, refractile cells with a rapid growth rate, and within 3–4 days, it overtook the slow growing HNGC-1 cells. Cell line HNGC-2 developed from this clone. The sequential phenotypic changes observed during development of HNGC-2 from HNGC-1 are depicted in Figure 2B. This spontaneously immortalized cell line is in continuous culture and has crossed more than 400 passages. HNGC-2 cells are fast-growing and showed exponential growth and did not exhibit senescence in culture (Fig. 2Ca). The HNGC-2 cells also had predominant population of multinucleated giant cells (Fig. 2Cb). The giant cells were highly positive for CD133 (Fig. 2Cc). The soft agar assay, an in vitro correlate of transformation, showed development of large colonies with HNGC-2 cells (supplemental online data S3c).
Loss of Senescence and Enhanced Telomerase Activity in HNGC-2 Cells
For immortalization, cells must bypass replicative senescence and overcome crisis by expression of telomerase to counteract the progressive shortening of telomere repeat sequences observed during in vitro proliferation of adult stem cells. The HNGC-1 cells entered into a crisis phase around passages p28–p32, became senescent, and a spontaneous clone developed from the growth-arrested cells. We therefore determined the telomerase activity in HNGC-1 and HNGC-2 cells. Although the telomerase activity was absent in the presenescent HNGC-1 cells, an enhanced activity was observed in HNGC-2 cells, with a level almost comparable to levels in the telomerase-positive control cells (Fig. 2Da). Similar telomerase activity was maintained in HNGC-2 cells assayed across two widely different passages (p50 and p60). The bypassing of replicative senescence and ability of cells for continued cell divisions in the face of dysfunctional telomeres and deregulated DNA repair machinery must be driving the cells toward genomic instability and contributing to tumorigenesis.
HNGC-2 Cell Line Demonstrates Characteristics of Brain Tumor Stem Cells
Brain tumors have been reported to arise by malignant transformation of NSCs. Since HNGC-2 cells that showed features of transformed cells were obtained from HNGC-1 cells, which are nontumorigenic stem cells (HNGC-1) by spontaneous immortalization, we explored the possibility that the HNGC-2 was a cancer stem cell line. We therefore analyzed the HNGC-2 cells for stem cell characteristics and their potential to form tumors. The characteristics of stemness include the ability to form neurospheres, express neural stem cell markers, and multilineage differentiation with markers for astrocytes, neurons, or oligodendrocytes. The HNGC-2 cells cultured in serum-free medium with supplements formed neurospheres with proliferating cells in the periphery. The primary neurosphere development was rapid, and almost 60%–65% of the cells had the potential to form medium- to large-sized neurospheres (Fig. 3Aa). The size of the neurospheres increased with time and, within 10 days in culture, most of the cells formed large neurospheres. The large neurospheres contained approximately 1,000 cells and had attained a size of 61. 66 ± 3.51 μm (Fig. 3Ab). The primary neurospheres could be cultured, passaged, and developed into secondary neurospheres. The secondary neurospheres appeared morphologically different from primary neurospheres, contained a fewer number of cells compared with primary neurospheres, had a size of 30–35 ± 2.97 μm, and were not as compact as primary neurospheres (Fig. 3Ac). Both the primary and secondary neurospheres were intensely positive for the neural stem cell marker CD133 and nestin (Fig. 3Ad, 3Ae). Neurospheres may also be formed due to aggregation of cells growing in suspension. To rule out this possibility, neurosphere formation was also assayed by using a clonogenic population of HNGC-2 cells at a very low seeding density. The cells generated from the neurospheres of the HNGC-2 cell line showed immunoreactivity for a panel of neural stem cell markers comprised of CD133, nestin, bmi-1, musashi1, Sox2, and nucleostemin (Fig. 3B), indicating the presence of a neural stem cell population in HNGC-2. The adherent HNGC-2 cells, as well as the neurospheres, were scored for immunopositivity for CD133 by flow cytometry. There was a significant increase in both percent positivity for CD133 expression from 49% in adherent cells to 98% in neurospheres as well as a 400-fold increase in fluorescence intensity for CD133 expression, as shown in Figure 3C. This indicates that HNGC-2 is a brain tumor stem cell line that has a high percentage of CD13-positive subpopulation and development of HNGC-2 cells as neurospheres selected only for the CD133 cancer stem cell population (Fig. 3Bc). The HNGC-2 cells that did not form neurospheres remained as single cells or doublets that, when specifically separated from the neurosphere population, were not found to be positive for CD133. This indicates that the HNGC-2 cells as neurospheres showed enrichment for CD133 population, a marker for both neural stem cells and brain cancer stem cells.
HNGC-2 Was Highly Tumorigenic and Invasive
The HNGC-1 cells injected to the extent of 107 cells were nontumorigenic when injected in nude mice, whereas the HNGC-2 cells formed large multilobed tumors . Next, we analyzed the potential of HNGC-1 and HNGC-2 cells to form tumors within the brain by injecting GFP-transduced cells intracranially into nude mice. Implantation of up to 107 GFP-tagged HNGC-1 cells and their corresponding neurospheres failed to show engraftment in the mouse brain up to 2 months assayed by lack of GFP positivity in the brain tissue. This observation was further confirmed with the help of histochemical analysis of H&E-stained brain tumor sections. In a sharp contrast, all of the mice (n = 4) injected with as few as 1,000 adherent HNGC-2 cells, as well as their corresponding neurospheres, developed tumors in the brain detected within 6–8 days of implantation. The brain tissue sections stained with H&E, obtained from mice implanted with HNGC-2 adherent cells as well as neurospheres, showed tumors with characteristic glioblastoma features. These included presence of atypical nuclei showing high mitotic activity, vascularization, and ventricular invasion away from the injection site (Fig. 4Aa). Characteristically, the mice injected with neurospheres showed development of larger tumors as compared with the tumors obtained on implantation of their adherent counterparts with increased vasculature, as can be seen by the presence of blood vessels within the tumor tissue sections (Fig. 4Ab). The tumors obtained with HNGC-2 adherent cells showed a number of GFP-positive tumor foci (Fig. 4Ba) with a large number of GFP-positive cells embedded in the brain parenchyma (Fig. 4Bb). A few of the cells appeared to be dopaminergic neurons (Fig. 4Bc). The brain tumor sections could be stained by the NSC markers with human-specific antibodies against nestin (Fig. 4Ca) and CD133 (Fig. 4Cb) The tumor foci, as well as tumor cells, stained positive with anti-human GFAP antibody, indicating that HNGC-2 cells had the potential to recapitulate the phenotype of the patient's tumor, thereby confirming the cancer stem cell nature of HNGC-2 cells (Fig. 4Cc, 4Cd). The presence of GFAP-positive cells in the tissue being specifically derived from the injected cells was confirmed by the presence of GFP staining in the same sections positive for GFAP (Fig. 4Ce).
The neurosphere-generated tumors appeared more vigorous in their growth potential (Fig. 4Da) and showed areas of large GFP-positive foci infiltrating the brain tissue (Fig. 4Db). The tumors were highly proliferative, as seen by the presence of a large number of overgrown cells stained with 4,6-diamidino-2-phenylindole (DAPI) (Fig. 4Dc). The size of the tumor focus formed in the brain was larger in mice implanted with neurosphere tumors compared with mice injected with an identical number of adherent cells for the same time period in vivo. The neurosphere-derived tumors were infiltrative, showed enhanced tumor growth, and increased vascularity, as seen by the presence of blood vessels scattered within the tumor stroma in the tumor tissue sections stained with H&E and DAPI, as well as by the presence of blood vessels in the GFP-positive tumor tissue sections (Fig. 4Dd). A marked angiogenesis was predominantly seen with glomeruloid blood vessels characteristic of GBMs. The tumor sections stained intensely for neural stem cells markers nestin and CD133 (Fig. 4Ea, 4Eb) and were highly positive for GFAP (Fig. 4Ec). The mouse tumor was excised, and 1,000 cells were reinjected into nude mice. Within a week, the reinjected cells were capable of tumor formation, confirming the self-renewal capacity of the cells. The potential of the cells to form large malignant tumors with small cell numbers in a short span of time is a predominant feature of CSCs and supports our hypotheses that HNGC-2 cells are enriched with a tumor-initiating stem cell population.
Microsatellite Instability in HNGC-2
Genomic instability may present itself as alterations in the length of short repeat stretches of coding and noncoding DNA, resulting in microsatellite instability. Microsatellite instability was scored using ISSR-PCR using primers based on repetitive DNA sequences, anchored at the 3′-end with unique sequences to prevent slippage. We found that the band patterns obtained with the three different primer combinations were different in HNGC-1 and HNGC-2 cells. Although the band pattern differed only marginally between HNGC-1 and HNGC-2 cells with primer (CA)8RG, significant differences in number and intensity of bands were evident with (CA)8RY and (CG)4RY primers. With primer (CA)8RY, four bands were found altered in HNGC-2 in relation to HNGC-1, whereas with primer (CG)4RY, three major bands and two minor bands below 1 kilobase were seen changed in HNGC-2 relative to HNGC-1 DNA (Fig. 5A). Such variations in band patterns and intensity studied with ISSR-PCR reflected global changes at the genome level and creation of genome destabilization in HNGC-2 that could have contributed to microsatellite instability. These observations suggest that HNGC-2 may exhibit a mutator phenotype that could be largely a consequence of inactivating mutations in DNA damage repair genes.
Constitutive Activation of pATM and Loss of p53 in HNGC-2 Cells
With the high level of alterations seen at the genomic level in HNGC-2 cells using primers to microsatellite repeat DNA sequences using ISSR-PCR, we thought it would be important to analyze the activation of molecules involved in DNA damage response. During DNA damage response, the ATM kinase is one of the first kinases to be activated. The HNGC-1 cells, when assayed for activated ATM, showed nuclear expression in only 5% of the cells (Fig. 5B). In comparison, in HNGC-2 cells grown under the same conditions, many cells had a constitutively activated ATM kinase. Interestingly, few of the cells had the entire nucleus appearing as a threaded structure (Fig. 5B). Tumor suppressor gene p53 is an important substrate for ATM and is involved in G1/S arrest and serves as an important surveillance mechanism for maintaining genomic integrity. HNGC-1 cells showed nuclear expression of p53, phospho-p53, and p21 (Fig. 5B). In contrast, the HNGC-2 cells showed a loss of expression of p53 as well as p21. These findings suggest that total abrogation of G1 checkpoint control in a nonarrested, genomically unstable proliferating pool of cells in HNGC-2 might have evolved due to loss of these two important molecules.
HNGC-2 Cells Show a High Number of 53BP1- and γH2AX1-Positive Foci
Following break recognition, ATM is activated to phosphorylate γH2AX and 53BP1. γH2AX may be used as an indicator for DNA double-strand breaks (DSBs), as the number of γH2AX foci detected by immunofluorescence is quantitatively correlated with the number of DSBs. It is assumed that the nuclear foci mark individual sites of DNA damage, indicating the coalescence of many substrates at the DNA damage sites. The HNGC-1 cells showed a diffused nuclear staining for γH2AX1 and 53BP1 (Fig. 5C). In contrast, the HNGC-2 cells, besides displaying loss of both proteins, p53 and p21, important in cell cycle arrest pathway, showed a large number of γH2AX1-positive foci (Fig. 5D). The large number of γH2AX1 foci is indicative of a high level of genomic instability in HNGC-2 cells. ATM-mediated phosphorylation of histone γH2AX also resulted in a large increase in the concentration of 53BP1 in the chromatin region adjacent to the DSB (Fig. 5D). A comparative analysis of the number of H2AX1 foci seen in both the cell lines was remarkably different, with almost 80%–90% of HNGC-2 cells exhibiting approximately 250 or more H2AX1 foci in contrast to HNGC-1 cells, which showed not more that five foci per cell (Fig. 5E). The HNGC-2 neurospheres, when checked for the presence of molecules involved in DNA damage activation and checkpoint control, showed an activation profile similar to that of adherent cells. A large number of γH2AX1 foci were seen in a number of neurospheres, as shown in Figure 5Fa, with a large number of damage foci seen in each neurosphere (Fig. 5Fb–5Fd).
Aberrant Expression of DNA Repair and Checkpoint Proteins in HNGC-2
We examined the activation of downstream targets of ATM/ataxia-telangiectasia and Rad3 related (ATR) kinases involved in DNA damage repair and checkpoint control. Immunofluorescence studies revealed high expression of Chk2 and Brca1 followed by Smc1, Chk1, Mre11, and Rad52 as distinct foci in HNGC-2 cells (Fig. 6B). In contrast, in HNGC-1 cells, all of these molecules showed diffused nuclear staining with no DNA damage foci (Fig. 6A). Brca2 was not expressed in both HNGC-1 and HNGC-2 cells. The aurora kinase A involved in G2/M checkpoint was overexpressed in HNGC-2 cells, wherein a majority of cells showed bipolar, tripolar, tetrapolar, and even hexapolar spindles (Fig. 6B).
The Cells Show Activation of Notch and Its Transcriptional Targets
The survival of neural stem cells is now known to be regulated by the activation of Notch and Hes pathway. This pathway is presently believed to provide a switch that balances survival and differentiation of neural stem cells. We hence looked at the presence of activated Notch in both the nontumorigenic stem cells (HNGC-1) and the cancer stem cells (HNGC-2) using an antibody that recognizes the active Notch intracellular domain. Both the cells HNGC-1 and HNGC-2 were positive for the presence of activated Notch and showed intense and specific expression of Notch-1 NICD in the nucleus (Fig. 7). HNGC-2 cells with 94% cells displayed a high positivity toward Notch-1, as analyzed with fluorescence activated cell sorting (supplemental online data S4). Both HNGC-1 and HNGC-2 cells were found to express Notch-2 and Notch-3 as well, although comparatively a significant overexpression of Notch-1, -2, and -3 was noted in HNGC-2 cells (Fig. 7). The transcriptional regulators of Notch, the basic helix-loop-helix proteins Hes1, Hes2, and Hes4, were also activated in both cells HNGC-1 and HNGC-2 (Fig. 7).
We identified a cell population in the adult human glioma tissue that could be cultured long term in vitro over limited passages (HNGC-1) displaying properties of multipotential stem cells that include ability to self-renew, positivity for neural stem cell markers CD133 and nestin, and ability for multilineage differentiation. The HNGC-1 cells, being derived from a glioma tissue, are categorized as nontumorigenic stem cells to differentiate them from normal adult neural stem cells.
Spontaneous immortalization of these cells (HNGC-1) led to development of an immortal clonal population (HNGC-2) with a potential to produce high-grade and invasive brain tumors in immunodeficient mice. The identification of cells with features of NSCs from an adult glioma tissue is intriguing. We believe that these cells are a part of the regional normal stem cell population embedded within the adult glioma tissue. Gliomas, being heterogeneous tumors, contain a mixture of neurons and glia that represent normal cells that are trapped within the tumor matrix . The normal cells like fibroblasts, endothelial cells, and smooth muscle cells contributing as stromal components are interspersed due to tumor infiltration in the brain . It has been recently shown that neural stem and progenitor cells show extensive tropism in intracranial xenograft glioma models, distributing themselves to the tumor bed and migrating in juxtaposition to advancing tumor cells . Thus, brain tumors are composed not only of progeny cells coming from the tumor cell of origin, but also of recruited normal cells migrating into the tumor from the brain and, possibly, other cells. The reports that NSCs may be present in the brain tumor are based on the observations that the subependymal plate or subventricular zone contains “embryonal nests” that have NSCs or progenitor cells with potential for transformation . Recent work in animal models and primary gliomas suggests that malignant gliomas arise from neural progenitor cells, and the endogenous adult neuroectodermal stem cells are cells of origin for primary brain tumors . Most tumors are derived from a single cell that has been transformed into a cancer initiating cell (cancer stem cell) with a capacity to proliferate and form tumor in vivo. It is still debated whether cancer stem cells would be derived from self-renewing normal stem cells due to their altered proliferative pathways or from progenitor cells that have acquired the ability to self-renew as a result of oncogenic mutations .
Defining Brain Tumor Stem Cells
Spontaneous transformation in human cells is rare, as they have two control points that regulate their life span in vitro—senescence and crisis phase. In this study, we provide evidence to show that HNGC-1 cells around p28 senesce and enter a crisis phase with a few cells escaping crisis and undergoing immortalization and spontaneous transformation. Human cells immortalize at a low frequency and are resistant to spontaneous transformation. Since HNGC-2 arose by spontaneous immortalization of HNGC-1 that are nontumorigenic stem cells, the HNGC-2 represented the glioma stem cells. The HNGC-2 cells satisfied most of the features of glioma stem cells, which included neurosphere formation, expression of neural/cancer stem cell markers—CD133, Sox-2, musashi, and nestin, ability for multilineage but aberrant differentiation with many cells coexpressing both neuronal and glial markers, and ability of the CD133-positive population to show high tumorigenic and engraftment potential in the brain. The fraction of adherent cells that were not forming characteristic neurospheres were CD133-negative. A distinct CD133-positive population has been identified from brain tumors of both adults and children that contains brain-tumor-initiating cells. These brain-tumor-initiating cells are the most aggressive cells that possess the capacity for self-renewal and can generate tumors that are a phenocopy of the patient's tumor with as few as 100 CD133-positive cells [13, 27].
Next, we investigated the molecular mechanisms that led to the evolution of the highly malignant CSC line HNGC-2 from nontumorigenic HNGC-1 stem cells. It appears that the self-renewal properties of HNGC-1 allowed them to proliferate without differentiating, thereby creating a pool of self-renewing stem cells. In these stem cells, further mutations followed by creation of genomic instability would have created a pool of self-renewing highly neoplastic brain tumor stem cells. This argument has also been proposed for evolution of CSCs .
Our data with ISSR-PCR and activation of telomerase in HNGC-2 cells provide evidence about the existence of a high level of genomic instability in HNGC-2 cells. It has been proposed that cells with cancer stem cell properties re-emerge from established cell line populations through epigenetic reprogramming and selection of subpopulation of cells with genomic instability .
A High Number of DNA DSBs and a Constitutively Over-Activated DNA Damage Response Pathway Are Causal of the High Level of Genomic Instability in HNGC-2
The occurrence of a DNA damage response was ascertained by monitoring histone H2AX phosphorylation (H2AX), 53BP1 intracellular localization, Chk2 phosphorylation (on Thr 68), and p53 protein levels [30, 31]. The HNGC-2 cells showed a large number of γH2AX and 53BP1 foci, indicating the existence of numerous DNA double-stranded breaks. The double-stranded breaks could be due to a high-level of replicative and genomic stress and not due to the culture conditions during in vitro propagation of HNGC-2. Both HNGC-1 and HNGC-2 cultures were always maintained under serum-free conditions, and only cells in early passage after their evolution (i.e., p8–p10 for HNGC-1 and p30–p32 for HNGC-2) were used. The cells were regularly characterized during their use in the experimental studies.
It has been demonstrated that the number of γH2AX foci detected by immunofluorescence is quantitatively the same as that of DSBs, suggesting that γH2AX may be used as an indicator for double-stranded breaks [32, 33]. The DSBs lead to activation of ATM/ATR kinase that phosphorylates a number of key players in numerous damage response pathways . The histone H2AX, on generation of DSBs, becomes phosphorylated to a form γH2AX, which promotes the formation of nuclear foci concentrating proteins like 53BP1, MDC1, Mre11/Rad50/NBS1 complexes, and Brca1, important for checkpoint function or DSB repair . In this manner, γH2AX augments cellular responses to double-stranded breaks and contributes to the preservation of genome stability .
The HNGC-2 cells showed a high number of foci with each of the pATM substrates: 53BP1, Mre11, BRCA1, and smc1. We always observed that, in HNGC-2 cells, the 53BP1 foci colocalized with the nuclear foci containing γH2AX, which seemed to mark the actual sites of DNA damage. This may be required for amplification of DNA damage response and checking the entry of cells with damaged DNA into mitosis . A direct link has been offered for the role of 53BP1 to cancer, with the finding that 53BP1 is persistently localized into nuclear foci in tumor cell lines lacking p53 [38, 39]. The other downstream effectors activated by ATM and ATR are the checkpoint kinases Chk1 and Chk2 [40, 41]. They together constitute a kinase cascade that amplifies the DNA damage signal and phosphorylates multiple targets, which are substrates of either ATM or Chk2 or both . Most of these substrates are effectors of the DNA DSB checkpoint pathway. One of the substrates for Chk2 is p53. Chk2 phosphorylates the N-terminal activation domain of p53, which is required for rapidly spreading the checkpoint signal from localized sites of DNA damage to targets, and thereby regulates p53 in response to DSBs . HNGC-1 cells with an appropriate DNA response and checkpoint arrest pathway due to presence of both p53 and p21 exhibit error-free DNA synthesis.
A recent study with glioma stem cells has implicated defective checkpoint responses and abnormal checkpoint control leading to genomic instability and serving as a potential contributor to the transformation of normal cells into cancer stem cells. In their study, it was found that, in response to radiotherapy, it was only the CD133-positive population that could survive the radiation, as it possessed an activated DNA damage checkpoint response that prevented it from undergoing senescence or apoptosis and allowed it to repopulate the tumor .
Aberrant Overexpression of Notch in Cancer Stem Cells
Notch proteins are known to play a fundamental role in cell fate decisions including proliferation, differentiation, and apoptosis. Hence, they have been implicated in tumorigenesis and appear to function as either oncogenes or tumor suppressor proteins, depending on their cellular context . Furthermore, although Notch receptors signal primarily through the regulation of hairy enhancer of split (HES) and HES-related genes, they are known to crosstalk with other signaling pathways, including the epidermal growth factor and the mitogen-activated protein kinase pathways . Aberrantly activated Notch signaling has also been documented in lung, breast, salivary gland, and pancreatic carcinoma . In a recent study, it is shown that Notch signaling levels were higher in the stem-like cell fraction in medulloblastomas, as Notch blockade reduced the CD133-positive cell fraction almost fivefold and totally abolished the side population, suggesting that the loss of tumor-forming capacity could be due to the depletion of stem-like cells . Stem-like cells in brain tumors thus seem to be selectively vulnerable to agents inhibiting the Notch pathway. In our study, we observed an over-activation of Notch and its transcriptional targets in HNGC-2 cells that are prototypes of glioma-derived cancer stem cells.
We believe that the constitutive activation of ATM-H2AX/Chk2-p53 as well as ATR-H2AX/Chk1 DNA damage checkpoint pathways in HNGC-2 cells may be linked to loss of p53, as p53 is the downstream target of Chk2 and ATM. Although our correlative data do not allow us to establish cause and effect relationships, one intriguing possibility is whether activation of DNA damage checkpoint pathway selects for p53 mutation. p53 becomes inactivated, as is typical of many tumors and precancerous lesions, then cells with compromised genome integrity pathways survive inappropriately, and the accrual of oncogenic lesions can fuel the carcinogenic process. We hypothesize that the high level of genomic instability in HNGC-2 cells must be contributing to the cancer stem cell nature of HNGC-2. The observed DNA damage might reflect abnormalities in prereplication complex maturation and/or stalled or collapsed replication forks, which are known inducers of the ATR-H2AX/Chk1 cascade.
Our study highlights the importance of DNA damage elements and checkpoint kinases as key signal transducers within the complex network of genome integrity checkpoints in tumors arising due to cancer stem cells. Based on our data, we propose a model that depicts the activation of molecules involved in DNA damage response that might be responsible for immortalization of the nontumorigenic stem cells to cancer stem cells using the HNGC-1 and HNGC-2 culture system (Fig. 8). Designing strategies for modulation of the DNA damage response and checkpoint kinases would thus become imperative in deciding on the therapeutic approaches for cancer treatment. With such a high activation of Notch isoforms 1, 2, and 3 in HNGC-2 not reported to date with any of the known established glioma cell lines, this further enhances the usefulness of this cell system for designing a specific multiagent chemotherapeutic regimen for targeting the cancer stem cells in glioblastoma.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.
This research was funded by the Department of Biotechnology (DBT), New Delhi, India. We acknowledge the technical expertise of Dr. Avanti Golwilkar and Dr. Mahesh M. Mandolkar, Golwilkar Metropolis, Pune, India, in histopathological analysis and of Dr. V.S. Baburao, Institute of Immunohaematology, Mumbai, in karyotyping. We thank Mr. Shaikh and Mr. Inamdar from the Experimental Animal Facility, NCCS, Pune, for their expertise in animal experimentation. We acknowledge the research fellowships provided by the Council of Scientific and Industrial Research (CSIR), New Delhi, India, to S.C.T., G.R.P., and R.G.; S.C.T. and V.S. contributed equally to this work.