Characterization of Glioma Stem Cells Through Multiple Stem Cell Markers and Their Specific Sensitization to Double‐Strand Break‐Inducing Agents by Pharmacological Inhibition of Ataxia Telangiectasia Mutated Protein

Chemicals

KU‐55933 was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA, product code: sc‐202963). KU‐60019 was purchased from Selleck Chemicals (Houston, TX, USA, product code: S1570). Gö 6976 was purchased from Sigma‐Aldrich (St. Louis, MO, USA, product code: G1171). BML‐277 was purchased from Enzo Life Sciences (Lausen, Switzerland, product code: BML‐EI388).

Cell lines and culture conditions

BORRU and DR177 cell lines derived from two adult GBM. BORRU cells had been isolated from a 74‐year‐old male patient and they have been previously described (19). The newly established DR177 cell line was derived from a hemispheric tumor of a 68‐year‐old Caucasian woman. Tissue surplus to diagnostic requirements was processed for isolation of GSC, according to Raso et al (17) with modifications. Briefly, tissue was washed with Dulbecco's modified Eagle's minimal essential medium (DMEM, Gibco‐Invitrogen, Milan, Italy), minced and triturated and cells were seeded at a concentration of 100 000 per mL into tissue culture flasks coated with Matrigel (BD Biosciences, Franklin Lakes, NJ, USA). Cells were grown in a proliferation medium containing DMEM‐F12/NSA and B27 supplement without vitamin A (1:50; Gibco‐Invitrogen), recombinant human fibroblast growth factor (FGF‐2, 10 ng/mL, Peprotech, Rocky Hill, NJ, USA), recombinant human epidermal growth factor (EGF, 20 ng/mL, Peprotech). Medium was routinely changed twice a week. Under these conditions, the cells adhered and grew in flasks into a monolayer and they maintained intact self‐renewal capacity. In the absence of matrigel, cells predominantly grew forming suspended neurospheres. Removal of growth factors and addition of 10% fetal calf serum (FCS) to the proliferation medium, after approximately 2 weeks, induced GSC differentiation, giving acquisition of astrocytic/neuronal morphology and increasing expression of the differentiation markers synaptophysin (SYN) and glial fibrillary acidic protein (GFAP) (Table 1).

Morphological characterization and p53 status

Cells were cultured in the presence of growth factors (stem conditions) or FCS (differentiating conditions) for 14 days and photographed with an inverted microscope (Olympus Europa Holding, Hamburg, Germany) equipped with a 20× objective in transmitted light.

Mutational screening of the p53 tumor suppressor gene (OMIM # 191117) was performed on the entire coding sequence. Four contiguous amplicons ranging from 347 to 615 bp with high primer overlap were directly sequenced on cDNA using the CEQ DTCS‐Quick Start kit (Beckman Coulter, Fullerton, CA, USA) and subsequently analyzed with the CEQ2000XL DNA Analysis System (Beckmann Coulter). These four segments included a minimal promoter sequence beginning 47 bp upstream from the transcription start site (5′UTR). The results were compared with the functional database of the p53 protein (IARC TP53 database; http://www‐p53.iarc.fr).

Immunocytochemical and karyotypical characterization

Cytospins containing up to 5 × 10⁵ cells/slide were prepared and fixed with 4% paraformaldehyde and incubated with antibodies (Ab) as summarized in Table 1. Cells were stained with hematoxylin and eosin (H&E). Slides tested with polyclonal Ab were sequentially incubated with a rabbit anti‐mouse Ab (Dako, Glostrup, Denmark) and alkaline phosphatase anti alkaline phosphatase (APAAP) complex (Dako), for 30 minutes at room temperature (RT). Slides tested with monoclonal Ab were visualized by incubating with the Envision Labelled Polymer‐AP (Dako) for 30 minutes at RT. Alkaline‐phosphatase substrate (Fuchsin, Dako) was subsequently added as chromogen. Quantification of cells positive for a specific marker was carried out by counting all the stained cells within 20 randomly selected fields per specimen, and the percentage was calculated based on the total number of the counted nuclei. The karyotype was determined using conventional cytogenetic analysis as described (22). Two different passages were analyzed.

Flow cytometry

The reactivity of monoclonal anti‐Prom1 Ab was assessed by indirect immunofluorescence and flow cytometric analysis. In detail, 10⁴ adherent cells were harvested with 0.5 mM ethylenediaminetetraacetic acid (EDTA) in phosphate‐buffered saline (PBS), washed, suspended in buffer [2% bovine serum albumin (BSA) in PBS] and stained with anti‐Prom1 (clone: AC133; Miltenyi Biotech, Paris, France) followed by a phycoerythrin (PE)‐conjugated anti‐isotype‐specific goat anti‐mouse antiserum (Southern Biotechnology Associated, Birmingham, AL, USA) as second‐step reagent. The PE‐conjugated second reagent Ab alone was used as negative control and all samples were analyzed for single‐color cytofluorimetric analysis (FACSCalibur Becton Dickinson, Mountain View, CA, USA). Cells were gated by forward and side scatter parameters and results are expressed as logarithm of red fluorescence intensity (arbitrary units) vs. number of events. For each analysis, 10 000 events were counted.

For cell cycle analysis, 90 000 cells were plated in 35 mm dishes, exposed to 1 µM KU‐55933 for 30 minutes, irradiated with 5 Gy and incubated for 192 hours. Cells were then harvested by PBS‐EDTA treatment, washed with PBS, fixed in 70% ethanol/water and permeabilized with 0.0015% NP40, washed and stained with 50 µg/mL propidium iodide (PI) solution containing 100 units/mL RNase type A, 10 mM EDTA. Samples were run on CyAn ADP cytofluorimeter (Beckman‐Coulter, Brea, CA, USA) and analyzed by the ModFit 3.2 computer program (Verity Software House, Topsham, ME, USA).

Quantitative reverse transcription‐polymerase chain reaction (RT‐PCR)

The minimum information for publication of quantitative real‐time PCR experiments (MIQE) (6) is provided as follows. Total RNA was extracted from 10⁶ cells by using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA) combined with silica column purification system (Invitrogen, Carlsbad, CA, USA) to optimize the isolation of high‐quality RNA. Quantification and quality assurance were performed using the NanoDrop spectrophotometer (NanoDrop Technologies Wilmington, DE, USA) and the Agilent 2100 bioanalyzer (Agilent Technologies, Waldbronn, Germany), respectively. Double‐stranded cDNA synthesis was performed using a two‐step cDNA synthesis kit (Invitrogen, Carlsbad, CA, USA) with Oligo(dT) 20 priming. Expression of stemness genes was assessed by quantitative PCR (qPCR) using in‐house designed systems following a fine‐tuning procedure (18). Amplifications were carried out in singleplex runs on 25 µL using Express‐Sybr GreenER qPCR‐SuperMix Universal (Invitrogen, Carlsbad, CA, USA) and specific primers were developed targeting the following genes: Bmi1 (NM 005180), CD15 (NM002033), Musashi (Msi) 1 (NM 002442), Msi2 (NM 170721), Nanog (NM 024865), Nestin (NM 006617), Oct4 (NM 001159542) and Sox2 (NM 003106). Table 2 lists primers used for marker analysis. Prom1 analysis was performed as previously described (18). qPCR efficiencies of each system were calculated using the equation: E = 10^−1/slope and data were considered comparable when the difference between the efficiencies was <0.1 (16). The normalized fluorescent signal was automatically calculated using an algorithm that normalizes the reporter emission signal. The threshold value applied to the algorithm generating the threshold cycle (Ct) was set at 0.05 in all experiments. Nonfluorescent signals were generated by these assays when genomic DNA was used as substrate. The relative quantification of genes transcript was performed according to the comparative method (2^‐ΔΔCt, Applied Biosystems User Bulletin no. 2P/N 4303859) using the value emerged by the median of Pmk2 and B2M as the normalizer (Ctref) and BORRU under differentiative conditions (FCS) as tissue control (ΔCt_ref).

ATM assay

BORRU cells were seeded in 25 cm² flasks and grown under stem or differentiative (FCS) conditions. Cells were irradiated with 5 Gy ionizing radiation and extracted at the indicated times with Cell Extraction Buffer (Life Technologies, Milan, Italy) containing phosphatase inhibitors plus protease inhibitor cocktail (Sigma, St. Louis, MO, USA). Total proteins were separated by SDS‐PAGE on 15% polyacrylamide gels (60 µg per lane) and were transferred to nitrocellulose membranes using iBlot‐Dry Blotting System (Life Technologies) according to the manufacturer's instructions. The membranes were then incubated with mouse monoclonal antibodies against: p53 phospho‐serine 15 specific antibody (Cell Signaling Technology, Danvers, MA, USA), p53‐specific (BIOCI, Airasca, Italy) and β‐actin (Cell Signaling Technology). Peroxidase‐conjugated goat anti‐mouse antibody was used as secondary antibody (Santa Cruz Biotechnology, Heidelberg, Germany). Immune complexes were visualized with the use of a Supersignal West Pico Chemiluminescent Substrate (Euroclone, Milan, Italy) on autoradiography films according to the manufacturer's instructions.

Survival

1500–5000 cells were seeded in 96‐well plates. The day after, cells were preincubated for 30 minutes at 37°C in complete medium with either the indicated ATM inhibitors or solvent [dimethylsulphoxide (DMSO)] and then treated. Treatment with ionizing radiation was performed by a CIS Bio International IBL437C irradiator at a fluence of 0.14 Gy/second. Treatment with camptothecin (CPT), mitomycin C (MMC) and hydroxyurea (HU) was performed by addition of the drug diluted in complete stem medium to the wells. Cells were then incubated at 37°C for 192 hours with no medium change. Survival was evaluated by the In Vitro Toxicology Assay Kit MTT‐based (Sigma, St. Louis, MO, USA). DMSO concentration never exceeded 0.38% v/v.

Statistics

The two‐tailed “Student's”t‐test was used. The nonparametric Wilcoxon paired t‐test was used for analysis of qRT‐PCR data. Statistical significances (P < 0.05; P < 0.01) were indicated with one (★) and two (★★) stars, respectively. Statistical analysis was performed using the statistical software GraphPad Prism 5.01 for Windows.

Establishment and characterization of DR177 glioma cell line

The DR177 cell line was derived from the GBM of a 68‐year‐old Caucasian woman. The culture was continuously passaged in vitro for more than 1 year with no changes in its morphology, proliferation, differentiation patterns as well as subsphere‐forming capacity, thus confirming its long‐term self‐renewal ability. The cells were oval to spindle in shape displaying nuclei with coarse chromatin and eosinophilic cytoplasm (Figure 2A). Mitotic figures were easily seen. The immunocytochemical characterization of DR177 is shown in Table 1 and Figure 2B,C. The stem cell phenotype (cells grown under serum‐free conditions in the presence of EGF and FGF) was associated with high levels of nestin (Figure 2B), while GFAP was only expressed by the neurospheres external layer (Figure 2C). When DR177 cells were cultured under differentiative conditions (absence of EGF‐FGF and presence of FCS), an increase of GFAP (a marker of differentiation to the astroglial lineage) and SYN (a marker of synaptic vesicles plasticity) expression was seen (Table 1). To assess the karyotype of the DR177, two passages were examined (3rd, 10th). The karyotype pattern remained stable in both stages, showing a modal chromosome number of 72, with either numerical or structural aberrations (Table 1).

Comparison of BORRU and DR177 glioma cell lines

BORRU cells derived from adult GBM have been previously described (19). Figure 3 shows a phenotypical comparison of BORRU and DR177 cells. Both BORRU and DR177 cells could be grown under serum‐free conditions in the presence of EGF and FGF (Figure 3A–D); moreover, they could be grown under either adherent conditions (provided culture vessels were matrigel coated—Figure 3A,B) or as suspension cultures in uncoated vessels where they formed relatively uniform round or oval neurosphere colonies (Figure 3C,D). Neither cell line harbored p53 mutations of the coding sequence associated with any loss of function (data not shown).

Figure 3E shows the relative gene expression of the stem cell markers. Expression of stem cell markers Prom1, Bmi1, CD15, Msi1, Msi2, Nanog, Nestin, Oct4 and Sox2 were determined for BORRU and DR177 cells by qPCR. All markers were expressed at a significantly higher level in BORRU as compared with DR177 cells. Hence BORRU cells displayed a stem cell marker expression phenotype that was significantly more pronounced as compared with DR177 cells. The Prom1 epitope (AC133) was further evaluated by flow cytometry (Figure 3F). The majority of BORRU cells (90%) harbored AC133 while it was present only in 1.33% of DR177. Figure 3G shows that DR177 cells grown under stem conditions (EGF‐FGF) proliferated faster than BORRU cells, especially from 4 to 8 days after seeding. Removal of growth factors and supplementation with 10% FCS resulted in differentiation of BORRU cells after approximately 2 weeks. Under those conditions, BORRU cells acquired astrocytic/neuronal morphology (Figure 3H), the expression of stem markers including Prom1 was ablated (Figure 3E) and their proliferation was arrested (Figure 3G), thus indicating that BORRU cells possess differentiation capacity. The latter was lower in DR177 cells cultured under the same conditions. Few DR177 cells acquired an astrocytic/neuronal morphology, with most of them remaining polygonal or fusiform (Figure 3I), the low expression of stem markers including Prom1 was ablated (Figure 3E) and proliferation was slowed down but not arrested (Figure 3G). Taken together, the above data indicate that DR177 cells possess a less pronounced stem phenotype as compared with BORRU cells.

Specific sensitization of BORRU cells to ionizing radiation and CPT by ATM inhibition

In the DNA damage checkpoint response, DNA damage is detected by the MRN/hSSB1 complex that activates the ATM and/or the ATR proteins (Figure 1). In turn, ATM/ATR unleashes an activation cascade that eventually culminates in cell cycle arrest at different phase passages. We (19) and others (3) have previously observed that the ATM substrates Chk1 and Chk2 are constitutively activated in GSC. In the present study, specific sensitization of BORRU, but not DR177, cells could be achieved by pharmacological inhibition of ATM protein.

2‐morpholin‐4‐yl‐6‐thianthren‐1‐yl‐pyran‐4‐one (KU‐55933) is an ATP‐competitive inhibitor that specifically targets ATM as compared with other phosphatidylinositol 3′‐kinase‐like kinases (Figure 1) (12). Cellular inhibition of ATM by KU‐55933 results in ablation of ionizing radiation‐dependent phosphorylation of a range of ATM targets, including p53, H2AX and Nijmegen breakage syndrome (NBS) 1 protein. The drug has low toxicity up to 10 µM on in vitro cultured GSC (data not shown). We exposed BORRU to 1 µM KU‐55933 30 minutes prior to ionizing radiation treatment and determined survival 192 hours later (Figure 4A). A limited but significant sensitization of BORRU was observed in the presence of KU‐55933, with D₃₇ dropping from 6.4 to 4.2 Gy. Analysis of the cell cycle distribution showed that a significant fraction of irradiated Go/G1 BORRU cells was pushed into G2/M phase after exposure to 1 µM KU‐55933 (P < 0.05; Figure 4C). A limited increase of DR177 cell survival was observed in parallel experiments, with D₃₇ values rising from 7.8 to >10.0 Gy in the absence or presence of KU‐55933, respectively (Figure 4B). No significant variation of cell cycle distribution was observed in DR177 cells exposed to the drug (Figure 4D). We then wished to determine whether the sensitizing effect could be specific of cells with stem properties. We exposed BORRU cells to differentiating conditions (removal of EGF‐FGF and addition of 10% FCS to growth medium for 14 days) (Figure 4E). Under those conditions, the sensitizing effect of 1 µM KU‐55933 was lost and resistance to ionizing radiation increased, with >90% surviving cells after irradiation with 10 Gy, independently of the presence or absence of KU‐55933 (D₃₇: >10 Gy in either case). No effect of KU‐55933 was observed on differentiated DR177 cells whose resistance to ionizing radiation slightly increased as a likely consequence of partial loss of proliferating capacity in comparison with stem culture conditions (D₃₇: >10 Gy in the absence or presence of KU‐55933, respectively; 3, 4). Hence, the above experiments indicate that the sensitizing effect of KU‐55933 is specific for BORRU cells grown under stem conditions.

The ATM inhibitor 2‐[(2R, 6S)‐2, 6‐dimethylmorpholin‐4‐yl]‐N‐[5‐(6‐morpholin‐4‐yl‐4‐oxo‐4H‐pyran‐2‐yl)‐9H‐thioxanthen‐2‐yl]‐acetamide (KU‐60019) is a recently synthesized analogue of KU‐55933 with Ki and IC₅₀ values half of those of KU‐55933 (10). KU‐60019 has been reported to be 10‐fold more effective than KU‐55933 at blocking radiation‐induced phosphorylation of key ATM targets in the established human glioma cell lines U87 and U1242 (10). Like KU‐55933, KU‐60019 displayed low toxicity up to 10 µM on in vitro cultured GSC (data not shown). We confirmed that ionizing‐radiation‐induced phosphorylation of p53 was effectively reduced in BORRU cells in the presence of 1 µM KU‐60019. Wild‐type p53 was expressed at elevated constitutive levels in these cells (Figure 5A). BORRU cells were preincubated with 1 µM KU‐60019 for 30 minutes, irradiated with 5 Gy and the ATM‐dependent p53 phosphorylation at serine 15 was monitored by Western blot analysis (Figure 5A). The phosphorylation of p53 was ablated in the presence of KU‐60019 indicating effective ATM inhibition. Under those conditions, BORRU cells were significantly sensitized to ionizing radiation. Cells were preincubated with 1 µM KU‐60019 30 minutes prior to ionizing radiation treatment and survival was determined 192 hours later (Figure 4G). In comparison with KU‐55933, more pronounced sensitization of irradiated BORRU cells was observed in the presence of KU‐60019, with D₃₇ dropping from 6.4 to 3.2 Gy. Symmetrically, the protecting effect on DR177 was more pronounced in the presence of KU‐60019 than in the presence of KU‐55933 (Figure 4H). The expression of stem cell markers in radiation‐survived BORRU cells was determined by RT‐PCR (Figure 5B). A significant decrease of stem markers expression was observed in cells survived 192 hours after exposure to KU‐60019 and irradiation with 5 Gy [BORRU (EGF/FGF) + KU‐60019 + 5 Gy] as compared with controls [BORRU (EGF/FGF)], supporting the hypothesis that the radiation treatment performed in the presence of KU‐60019 preferentially kills the stem‐like cells.

We further determined whether ATM inhibition could preferentially sensitize GSC to the DNA topoisomerase I inhibitor CPT that likewise ionizes radiation, induces significant numbers of DNA double‐strand break (DSB). Cells were preincubated with 1 µM KU‐60019 30 minutes prior to CPT treatment and survival was determined 192 hours later. Significant sensitization to CPT was observed in BORRU in the presence of 1 µM KU‐60019, with D₃₇ dropping from >25 nM to 19 nM (Figure 6A) but not in DR177 (D₃₇ of 15 and 13 nM in control and KU‐60019‐treated cells, respectively—Figure 6B). On the contrary, no sensitization of BORRU cells to other types of DNA damage/stress such as interstrand DNA cross‐links induced by mitomycin C (MMC) or replication stress induced by hydroxyurea (HU) was observed (Figure 6C–F), indicating that this phenomenon may be specific for agents inducing DSB.

In order to investigate the mechanism of BORRU sensitization following ATM inhibiton, we inhibited Chk1 and Chk2 that are downstream effectors of ATM in the DNA damage response (Figure 1). Exposure to the Chk1 inhibitor Gö 6976 (0.25 µM) resulted in sensitization of both BORRU and DR177 cells to ionizing radiation, with D₃₇ dropping from 6.4 to 3.4 Gy and from 7.8 to 3.7 Gy, respectively (Figure 7A,B). On the contrary, exposure to the Chk2 inhibitor BML‐277 (3 µM) resulted in increased survival of both BORRU and DR177 cells after ionizing radiation with D₃₇ rising from 6.4 to >10 Gy and from 7.8 to >10 Gy, respectively (Figure 7C,D). These results suggest that sensitization by ATM inhibitors may at least partially be linked to fading of Chk1 activity.