Background
According to the tumor stem cell hypothesis, resistance to conventional therapies may reside in a subset of tumor cells with stem-like characteristics [
1‐
3]. These cells are called cancer stem cells (CSCs) or cancer stem-like cells and are endowed with long-term self-renewal and a certain differentiation capacity. Several reports suggest that CSCs are indeed more resistant to standard chemo- and radiation therapy than non-CSCs [
4‐
13]. However, most studies addressing cell death modalities have focused on apoptosis early after the genotoxic insult [
6,
9‐
12]. The importance of mitotic catastrophe as cause of cell death induced by genotoxic treatments has so far not been addressed in CSCs. Mitotic catastrophe is caused by altered mitoses and/or irreparable chromosome damage and is accompanied by micronucleation and multinucleation. Mitotic catastrophe causes a delayed mitosis-linked cell death and finally leads to apoptosis or necrosis [
14‐
17].
Several explanations have been proposed for the higher gamma (γ)-ionizing radiation (IR) resistance of CSCs compared to non-CSCs: a stronger activation of DNA damage checkpoints associated with more proficient DNA damage repair [
6], less initial DNA damage due to lower levels of γIR-induced oxidative radicals [
7,
13], as well as activation of stemness pathways [
7,
8]. However, compared to conventional glioblastoma cell lines, glioblastoma CSCs were either more radiosensitive and repaired γIR-induced DNA-double strand breaks (DSBs) less efficiently [
18] or showed no difference in radio- and chemotherapy-induced DNA damage and repair [
19,
20]. Thus, the differences between CSCs and non-CSCs in γIR-induced DNA damage, damage repair and cell death are not fully clear.
We established cultures of immature stem-like cells from primary glioblastomas. Removal of the stem cell culture cytokines epidermal growth factor (EGF) and fibroblast growth factor-2 (FGF-2) and addition of fetal bovine serum (FBS) led in some but not all cases to differentiation of these stem-like cells. Using such directly related cultures, we examined the radioresponse of stem-like glioma cells (SLGCs) and of more differentiated glioma cells in terms of cell death mechanisms, focusing on both apoptosis and mitotic catastrophe. We also assessed whether the stem cell culture cytokines EGF and FGF-2 contribute to differences between stem-like and more differentiated tumor cells in terms of DNA damage levels and of apoptosis resistance upon γ-irradiation.
Materials and methods
Tumor samples and cell culture
Brain tumor samples were obtained following approval by the University of Freiburg ethical board (application number: 349/08) and informed written consent of patients. All patients were diagnosed as classical primary GBM. Tumors were dissociated into single cells with "Liberase Blendzymes" (Roche) for 45 min at 37°C. Cells were then allowed to form spheres in suspension culture in serum-free Neurobasal medium (Gibco) supplemented with EGF/FGF-2 (20 ng/ml each), B27, non-essential amino acids, penicillin/streptomycin, glutamax and heparin, on low attachment plates (Corning). For experiments, the cultures were expanded in plates coated with ECM proteins (mouse sarcoma-derived ECM, Sigma). The CSC-like properties were confirmed with serial neurosphere assays and serial xenotransplantation assays in BALB/c nude or non-obese diabetic/severe combined immunodeficient mice which were performed in accordance with protocols specifically approved by the animal care committee of the Regierungspräsidium Freiburg (registration number: G-10/64). Two SLGC cultures (G179 and G166) have previously been described by Pollard
et al. [
21] and were purchased from Biorep (Milan, Italy). For differentiation, the SLGCs were either transferred to DMEM supplemented with 10% FCS, penicillin and streptomycin, L-glutamine, non-essential amino acids and β-mercaptoethanol or to Neurobasal medium without EGF and FGF, supplemented with all-
trans-retinoic acid (Sigma).
γ-irradiation
Irradiations were performed using a Gammacell 40 137Cs laboratory irradiator.
Cell Growth and Viability Assay
An aliquot of cell suspension was mixed with Trypan blue solution (0.4% in PBS; Sigma), and the numbers of live and dead cells (viable cells excluded the dye and were unstained, nonviable cells were blue) were counted under a microscope.
Apoptosis assays
Exponentially growing cells that had been seeded 24-60 h before were irradiated, and at the time points indicated stained with Annexin V and propidium iodide (PI) using an Annexin V-FITC Kit from Milteniy Biotec. Apoptosis was measured by flow cytometry on a Cytomics FC 500 instrument from Beckman Coulter.
Assessment of mitotic catastrophe
24 to 48 h after seeding, cells were irradiated and, at the time points indicated, fixed and stained with 4'-6-diamidino-2-phenylindole (DAPI) for chromosome analysis under an Olympus BX41 fluorescence microscope equipped with a digital camera CC-12 soft imaging system (U-CMAD3, Olympus). For each assessment of the extent of mitotic catastrophe 200 nuclei were examined.
Immunofluorescence staining
Cells grown on slides were fixed with Histofix for 15 min at room temperature. Thereafter, the cells were permeabilized with 0.2% Triton-X100. After blocking (with 2% bovine serum albumin and 5% goat serum in PBS for 1 h at room temperature), the cells were incubated with primary antibodies against one of the following proteins: Sox2 (Abcam), CD133 (Milteniy), GFAP (Dako), nestin, Tuj, or musashi (Chemicon) at 4°C for 1 h or overnight, followed by incubation with Alexa Fluor 488-labeled secondary antibodies (Invitrogen) for 20 min at room temperature. Nuclei were counterstained with DAPI, and cells analyzed using a BX41 fluorescence microscope (Olympus) equipped with the digital camera CC-12 soft imaging system U-CMAD3 at 100-fold magnification. CD133+ cells were isolated from CSC cultures with magnetic beads coated with CD133 antibody (Milteniy).
Western blot analyses
Cell lysates were prepared in RIPA lysis buffer supplemented with protease inhibitor cocktail (Complete from Roche) and phosphatase inhibitors NaF and 7 Na3VO4 (Sigma). The blots were probed with the indicated antibodies and developed by enhanced chemiluminescence (Amersham Biosciences). The following antibodies were used: Sox2 (Abcam), musashi (Chemicon), nestin, γH2AX, p53, phospho-p53, Bcl-2, Bcl-xL, Mcl-1 and p21 from Cell Signaling, DNA-PK (BD Pharmingen), phospho-DNA-PK (Abcam), as well as actin and Bax (Santa Cruz). Quantification of signals was performed using Image Quant TL (Amersham Bioscience).
Blocking the EGF and the FGF2 pathway
The binding of cytokines was blocked at the receptor level with monoclonal antibodies. The anti-EGFR antibody Cetuximab (Erbitux®; Merck KGaA, Darmstadt) was used at a concentration of 60 nM and the anti-FGFR1 monoclonal antibody (clone VBS1, Chemicon) at a concentration of 5 μg/ml. The antibodies were added 1 h prior to adding the cytokines.
Cell surface marker determination by flow cytometry
Directly-PE-labeled antibodies against an extracellular glycosylation-dependent epitope (AC133) of CD133 (Milteniy) were used.
Cell cycle analyses
Exponentially growing cells seeded 60 h before were irradiated, fixed at the indicated time points with 70% ethanol, and stored overnight at -20°C. Cells were then washed and incubated with PI (50 μg/mL) and RNase (100 μg/mL) for 2 h at 4°C. After washing, the cells were analyzed for DNA content by flow cytometry.
Statistical analyses
All data are presented as mean ± SD and analyzed by Student's t test, two-tailed, with unequal variance. P < 0.05 was considered significant.
Discussion
Glioblastomas are among the most aggressive tumors, being incurable to date [
36] and CSCs are thought to have a major impact on their therapy resistance [
4‐
6]. Independent of tumor entity, CSCs are generally thought to be relatively radio- and chemotherapy-resistant, but, regarding cell death modalities, investigations have thus far concentrated on apoptosis occurring early after the genotoxic insult [
6,
9,
10,
12]. We confirmed here that primary SLGCs are extraordinarily resistant to apoptosis within the first few days after γ-irradiation (up to 96 h and up to doses as high as 10 Gy in this study), but we also show that at such doses, SLGCs very late (>4 days after irradiation) can undergo apoptosis. However, this late apoptosis was restricted to SLGCs undergoing G2M arrest resulting in mitotic catastrophe and seems thus to be restricted to proliferating cells. All these SLGC lines had nonfunctional p53. However, one line with nonfunctional p53 which underwent IR-induced G2M arrest did not undergo mitotic catastrophe or late apoptosis. Mitotic catastrophe thus seems to be an absolute requirement for this late type of cell death. FBS-differentiated cells tended to show significant apoptosis already between 48 and 96 h post-irradiation. Differences in apoptosis early after irradiation did not strictly correlate with observed differences in initial or residual γH2AX. In addition, the γH2AX levels turned out to depend strongly on the presence of the CSC-culture cytokines EGF plus FGF, confounding
in vitro analyses of DNA damage and repair.
We have for the first time compared radiation responses of primary stem-like glioma cells with that of more differentiated cells directly derived from the former. So far, similar studies have either compared CSC surface marker-positive with surface marker-negative cells [
6,
8,
11,
13], traditional cell lines cultured as spheres or adherently in FBS [
7], or else patient-derived CSC-like cells with traditional FBS-cultured cell lines [
18,
19]. In our study, only one SLGC line displayed substantial surface AC133 (the stem cell-specific glycosylation-dependent epitope of CD133). However, there is considerable evidence for AC133/CD133 negative tumor stem cells in glioblastoma [
23,
24].
It is generally assumed that differentiated tumor cells are more susceptible to genotoxic treatment-induced apoptosis than stem-like tumor cells. In our study, three SLGC lines (GBM8, G179, and G166) could be induced to differentiate by withdrawing EGF and FGF-2 and adding FBS. Although all these differentiated lines underwent more γIR-induced apoptosis than the corresponding SLGC cultures, a large increase was only observed for the FBS culture of one line, GBM8. An explanation for this might be that GBM8 SLGCs differentiate more readily than G179 and G166 SLGCs. GBM8 SLGCs lose Sox2 expression completely already a few days after EGF/FGF removal (not shown), whereas this process is much slower in the case of G179 and particularly G166 where the maximal (and incomplete) decrease of Sox2 expression was observed only after 3 to 4 weeks. Thus there may well be differences in the degree of differentiation among these three differentiated lines, but other mechanisms explaining the difference in the apoptotic response are also conceivable.
Cell death caused by mitotic catastrophe can occur at the first cell division after irradiation or at one of the next thereafter either as secondary apoptosis or as necrosis [
14,
16]. Mitotic catastrophe can be enhanced in cells lacking p53. In contrast, p53 is crucial for genotoxically induced apoptosis in cell types prone to primary apoptosis [
15,
37‐
39]. We observed here that most SGLCs with nonfunctional p53 underwent apoptosis mostly later than 4 d post-irradiation, i.e., presumably postmitotic, secondary apoptosis. Why one of the SLGC lines with nonfunctional p53 and γIR-induced G2M arrest failed to undergo mitotic catastrophe and late apoptosis is unclear at the moment and will be the subject of further studies in our laboratory.
Many primary glioblastomas are wild-type for p53 [
40]. However, in line with previous studies on primary GBM [
41] we found that p53 is not functional in most (4/6) primary glioblastomas studied by us. A possible reason for the lacking p53 stabilization is the deletions or mutations in PTEN frequently associated with primary GBM, since PTEN has been described as important for p53 stabilization [
10,
42]. Other possible causes are mdm2 gene amplification, loss of p14
ARF[
43] or overexpression of the NF-κB-signaling component RIP1 [
44], all frequently associated with GBM. Interestingly, the γIR-induced phosphorylation of p53 at serine 15 was intact, presumably because of an intact ataxia teleangiectasia-mutated pathway.
The non-differentiating FBS cultures of GBM4, GBM22 and GBM10 were almost completely apoptosis-resistant in the first 4 d after irradiation despite the prolonged γH2AX signals and the week-long absence of EGF and FGF-2. This early apoptosis resistance thus seems to be independent of potential direct anti-apoptotic effects of EGF and FGF-2 [
31‐
34]. In line with this, Bao
et al. described that, 20 h after irradiation, CD133+ glioblastoma cells showed apoptosis resistance irrespective of whether EGF and FGF were present after irradiation [
6]. However, immediately after cytokine withdrawal, anti-apoptotic and DNA-repair proteins directly induced by EGF/FGF may still be operating as we have observed differences in γH2AX after cytokine withdrawal for 72 h but not for 16 h. Thus, early after cytokine withdrawal, direct cytokine effects and stem cell-intrinsic effects on cell survival and DNA damage signals are not yet discernible.
It has so far not been studied whether EGF and FGF-2 confound comparisons of DNA damage signals between stem-like and non-stem-like cells. We observed that acute EGF/FGF addition to differentiated tumor cells reduces IR-induced γH2AX-signals while subacute and chronic withdrawal from non-differentiating stem-like cells increases them. This implies that, due to the different cytokine requirements,
in vitro studies on previously suggested intrinsic differences in γIR-induced DNA-damage [
7,
13] and repair [
6] between CSCs and non-CSCs are problematic. Thus, a strict
in vivo comparison would be best, but comprehensive experiments on precisely defined cell populations will hardly be possible. Very recently, two other groups reported that EGF [
45] and FGF-2 [
46] activate DNA damage/repair as reflected by reduced γIR-induced γH2AX signals. In these studies, conventional bronchial carcinoma cell lines [
45] or primary keratinocyte progenitor cells [
46] were used. The results from these studies support our findings.
We have observed that acute EGF/FGF-2 addition to differentiated cultures of p53-deficient SLGCs even enhanced γIR-induced apoptosis rather than decreasing it, despite the usually assumed pro-survival effects of EGF and FGF-2 and even though the γH2AX signals were reduced by the addition of the two cytokines. However, proliferation was accelerated, indicating a stronger correlation of the observed delayed γIR-induced apoptosis to cellular proliferation than to DNA damage assessed by γH2AX. We similarly examined two p53-deficient SLGC lines and, depending on the concentration of the two stem-cell mitogens, there was a correlation of cellular proliferation with mitotic catastrophe and late apoptosis induced by 5 or 10 Gy irradiation. However, this observation may only apply to tumor lines whose postirradiation survival is independent of EGF and FGF.
Mitotic catastrophe is associated with proliferation. It is often argued that CSCs, like many types of normal tissue stem cells, may be relatively quiescent, but it has been shown that even some normal tissue stem cells proliferate very fast
in vivo[
47]. Moreover, a recent study showed that in some GBM patients CD133+ glioblastoma cells coexpress the proliferation marker Ki67 [
48]. In addition, in cultures of normal neural stem cells, CD133 is expressed on the surface of proliferating cells [
49]. We have made similar observations in the CD133+ SLGC line examined by us, and accordingly found even more IR-induced mitotic catastrophe among bead-purified CD133+ cells than in the CD133- immature cells of the same tumor. It is however unclear at the moment whether this observation can be generalized.
Recent publications showed that transient pretreatment with a proliferation-inducing agent leads to the effective chemotherapy-mediated elimination of dormant hematopoietic stem cells and human primary leukemia stem cells
in vivo[
50,
51], but so far it has not been determined whether this also applies to genotoxically treated solid tumor stem cells. Our results suggest that this might only apply to those CSCs which undergo genotoxic treatment-induced mitotic catastrophe. Besides a search for suitable proliferation-inducing agents, more research is necessary regarding the proliferation status of CSCs and cell death associated with mitotic catastrophe in cancer patients undergoing genotoxic therapies.
Competing interests
The authors declare that they have no competing interests. NE is an employee of Proquinase, but Proquinase did not sponsor the study.
Authors' contributions
EF performed most of the experiments and contributed to writing the manuscript. SG, CT, and NE performed experiments. AW collected the tumor material and provided clinical information. GN was the Principal Investigator of the study and largely wrote the manuscript. All authors read and approved the final manuscript.