Cancer Cell Biology
STAT3 is essential for the maintenance of neurosphere-initiating tumor cells in patients with glioblastomas: A potential for targeted therapy?
Article first published online: 19 APR 2010
Copyright © 2010 UICC
International Journal of Cancer
Volume 128, Issue 4, pages 826–838, 15 February 2011
How to Cite
Villalva, C., Martin-Lannerée, S., Cortes, U., Dkhissi, F., Wager, M., Le Corf, A., Tourani, J.-M., Dusanter-Fourt, I., Turhan, A. G. and Karayan-Tapon, L. (2011), STAT3 is essential for the maintenance of neurosphere-initiating tumor cells in patients with glioblastomas: A potential for targeted therapy?. Int. J. Cancer, 128: 826–838. doi: 10.1002/ijc.25416
- Issue published online: 19 APR 2010
- Article first published online: 19 APR 2010
- Accepted manuscript online: 19 APR 2010 12:00AM EST
- Manuscript Accepted: 30 MAR 2010
- Manuscript Received: 31 AUG 2009
- La Ligue contre le Cancer de la Vienne et des Deux-Sèvres
- Sport et Collection
- Rotary Club de Civray
- Institut National du Cancer
- stem-like cells;
- Top of page
- Material and Methods
- Supporting Information
Glioblastoma (GBM), the highest-grade form of gliomas, is the most frequent and the most aggressive. Recently, a subpopulation of cells with stem cells characteristics, commonly named “tumor-initiating stem cells” (TISCs) or “cancer stem cells” (CSCs) were identified in GBM. These cells were shown to be highly resistant to chemotherapeutic drugs and to ionizing radiations. Consequently, the knowledge of the signals that regulate the functions and survival of TISCs is crucial. In our work, we describe a neurosphere-initiating cell (NS-IC) assay to quantify TISC/CSCs from patients with GBM and show that these cells are tumorigenic in vivo. We demonstrate that the intracellular signal transducer and activator of transcription STAT3 is constitutively activated by phosphorylation preferentially on serine 727 in these cells. Moreover, we demonstrate that the selective inhibition of STAT3 by the chemical compound Stattic or by siRNA STAT3 abrogates TISC/CSC proliferation and NS-IC suggesting that self-renewal of GBM “stem-like” cells depends on the presence of STAT3 for their maintenance. Finally, we show that inhibition of STAT3 by Stattic sensitizes TISC/CSCs to the inhibitory action of Temozolomide with a strong synergistic effect of both drugs. Overall, these results suggest that strategies focused on STAT3 inhibition are efficient at the level of “stem-like” cells and could be of interest for therapeutic purposes in patients with malignant GBM.
Glioblastoma (GBM), the highest grade form of gliomas (grade IV in WHO classification) is the most frequent and the most aggressive brain tumor. The current standard of care for patients with GBM includes tumor resection followed by adjuvant radiation therapy (RT) and chemotherapy (CT). A survival benefit was reported for GBM patients treated with Temozolomide (TMZ) combined to RT.1 Despite advances in surgical and medical neurooncology, the prognosis of GBM remains poor and the median survival is 14.8 months.2, 3 Because of its infiltrating characteristics, the complete resection of GBM remains impossible and the tumor recurrence occurs invariably at the primary localization of the tumor. These recurrences seemed to be due to the resistance of the initial tumor but recent data suggest that they may be linked to the presence in the tumor of a subpopulation of cells with stem cells characteristics, also called “tumor-initiating stem cells” (TISC) or “cancer stem cells” (CSC). GBMs are the second type of solid cancers in which tumor cells with stem cell-like features were identified,4–8 after breast cancer.9In vitro, these cells grow as spheres named neurospheres and they could recapitulate the parent tumor when transplanted into the rodent central nervous system and retain tumor forming capacity through serial transplantation. First reports suggested that tumorigenic cells in GBM were restricted to the CD133+ population4–6; however, recent studies showed that CD133− cells isolated from GBM are also tumorigenic.10, 11 Moreover, it has been suggested that GBM stem cells and progenitor cells are highly resistant to conventional chemotherapeutic drugs, including TMZ.12–14 In addition, these cells are radioresistant and could be the cause of tumor recurrence after RT.15 The chemo/radio-resistance of such tumorigenic cells may be responsible for the poor clinical outcome of GBM patients. Consequently, it may be essential to target TISC/CSCs and therefore the knowledge of the signals regulating the functions and survival of TISC/CSCs is crucial.
Several studies have identified key players involved in the intracellular signal transduction pathways regulating stem cell renewal and proliferation in neural cancer cells such as Wnt, Bmi-1 and Shh pathways.16–18 The intracellular signal transducer and activator of transcription STAT3 is a crucial self-renewal factor of murine ES cells.19, 20 This transcription factor is activated through tyrosine and serine phosphorylation by a wide range of normal antiapoptotic or proproliferative extracellular messages as well as by multiple oncogenic stimuli.21, 22 STAT3 is activated in many murine and human malignant tumors, including primary GBMs and glioma cell lines23–26 and recent studies reported that inhibition of STAT3-mediated signaling pathway reduces growth of GBMs cell lines.27–32 However, its role in the maintenance of tumor stem cells is still an open question.
In our work, we show that STAT3 is constitutively activated at the level of primary neurospheres from patients with GBM. By developing a quantitative neurosphere-initiating cell (NS-IC) assay, we demonstrate that the selective inhibition of STAT3 can abrogate neurosphere growth and self-renewal, suggesting that this selective targeting would be valuable to inhibit tumor growth at the level of primitive cells.
Material and Methods
- Top of page
- Material and Methods
- Supporting Information
Primary tumor sphere culture
Tumor samples were obtained within 30 min after surgical resection from two adult GBM patients (CSC-GBM1 and CSC-GBM2). Written informed consent forms were obtained from both patients. Tumor tissue was washed and mechanically dissociated. Cells were resuspended in a defined serum-free medium (DMEM/F12 or NBE, Invitrogen, Carlsbad, CA, http://www.invitrogen.com) containing 20 ng/mL of basic fibroblast growth factor (bFGF, Invitrogen), 20 ng/mL of epidermal growth factor (EGF, Invitrogen) and the culture supplements N2 (100×, Invitrogen) and B27 (50×, Invitrogen). Cells were plated at a density of 106 cells per 25 cm2 flask. As a control, the U-251MG human glioma cell line (provided by Dr. Mesnil, UMR CNRS 6187) was grown in the same conditions as the primary cultures.
Culture medium was replaced twice a week. When primary spheres became large and numerous, they were enzymatically dissociated with accutase (Chemicon International, Temecula, CA, http://www.chemicon.com) into single cells and their capacity to generate secondary or tertiary neurospheres was tested. To determine their differentiative capacity, neurospheres were grown in differentiation promoting conditions in the presence of DMEM high-glucose medium supplemented with 10% of fetal bovine serum (Invitrogen).
Evaluation of tumorigenicity by orthotopic injection
Tumorigenicity was determined by injecting GBM multiforme-derived neural stem-like cells (CSC-GBM1 and CSC-GBM-2) orthotopically into nude mice. Cell suspensions (105 and 103 cells in 10 μL of PBS) were delivered into the right striatum by stereotactic injection using a Hamilton syringe. Mice were sacrificed in view of the first clinical signs (weight loss). Hematoxylin and eosin staining were performed on 15-μm thick cryostat section. All experiments were done in accordance with the institutional animal care protocols.
Cell proliferation/survival assays
To determine the doubling times of neurospheres from tumors and cell lines, as well as the inhibitory effect of shRNA-STAT3 on proliferation, cell numbers were monitored for 20–30 days. The XTT cell proliferation test (Roche, Basel, Switzerland, http://www.roche.com) was used to determine the effect of Stattic (Chemicon) and TMZ (Interchim, Montluçon, France, http://www.interchim.com) on proliferation. Cells were plated in 96-well plates at a density of 5 × 104 cells per well and the quantification of the viable cells was performed on a MRXII (Dynex Technologies, Chantilly, VA, http://www.dynextechnologies.com) at 450 nm. The 50% inhibitory concentrations (IC50) of both drugs were determined after treatment with increasing amounts for 96 hr. The inhibitory effect of drugs at IC50 on proliferation was measured at regular intervals for 12 days. To determine the efficacy of Stattic and TMZ combinations, cells were treated with either IC20 of Stattic (0.5 and 1.07 μM for CSC-GBM1 and CSC-GBM2, respectively), or IC20 of TMZ (245 and 215 μM for CSC-GBM1 and CSC-GBM2, respectively), or in combination for 96 hr.
Clonogenic assays in methylcellulose
CSC-GBM cells (4 × 104 cells) were plated in 30-mm Petri dish (Greiner Bio One, Courtaboeuf, France, http://www.greinerbioone.com) in 1 mL of complete methylcellulose medium S1 (Stem Cell Technologies, Vancouver, BC, Canada, http://www.stemcell.com) and supplemented with B27 (50×, Invitrogen) and N2 supplements (100×, Invitrogen). Cells were treated with either IC20 of Stattic (0.5 and 1.07 μM for CSC-GBM1 and CSC-GBM2, respectively), or IC20 of TMZ (245 and 215 μM for CSC-GBM1 and CSC-GBM2, respectively), or in combination. After 20 days of incubation, colonies consisting of more than 50 cells were counted under inverted microscope. The experiment was repeated three times.
Annexin V staining and flow cytometry
Annexin V staining assays were performed by using the FITC Annexin V apoptosis detection kit (BD Biosciences, San Diego, CA, http://www.bdbiosciences.com) following the manufacturer's instructions. Briefly, aliquots of 105 cells were washed three times in phosphate-buffered saline (PBS) and resuspended in 100 μL of binding buffer (10 mM Hepes/NaOH (pH7.4), 140 mM NaCl, 2.5 mM CaCl2). A total of 5 μL of annexin V-FITC and 7 μL of 7-AAD (BD Biosciences, San Diego, CA, http://www.bdbiosciences.com) were added to test cells, followed by incubation at room temperature in the dark for 15 min. Cells were immediately analyzed by using FACSCalibur and CellQuest software (BD). A total of 30,000 events were analyzed in one typical experiment. Cells showing red fluorescence (7-AAD+) due to 7-AAD uptake were considered necrotic (i.e., cells had lost their membrane integrity). Only cells that were 7-AAD− and also stained with annexin V-FITC (annexin V+/7-AAD−) were considered apoptotic.
Glass discs were coated with polylysine overnight. Cells were plated on glass discs and fixed by covering the surface with a 4% paraformaldehyde solution for 15 min at room temperature. Slides were then washed with PBS and incubated in a permeabilization/saturation solution (0.5% Triton X-100/human serum albumin 1%) for 1 hr at room temperature. Cells were incubated overnight at 4°C with antibodies raised against nestin (1:200, Chemicon, Billerica, MA), SOX2 (1:100, Chemicon, Billerica, MA), CD133 (1:10, Miltenyi Biotec, Paris, France, http://www.miltenyibiotec.com), Stat-3/clone 124H6, pTyr705-Stat3/clone D3A7, pSer727-Stat3/clone 6E4 (1:250, Cell Signaling, Danvers, MA, http://www.cellsignal.com), GFAP (1:1000, Dako, Glostrup, Denmark, http://www.dako.com), β-tubulin III (1:250, Sigma, St Louis, MO http://www.sigmaaldrich.com), O4 (1:500, R&D Systems, Minneapolis, MN, http://www.rndsystems.com) or S100 (1:400, Dako). Alexa®555 and 488 labeled anti-IgG (1:200, Invitrogen) and TOPRO (1:1,000, Invitrogen) were used as secondary antibodies and nuclear marker. Slides were mounted with Prolong (Invitrogen) for viewing with a confocal laser scanning microscope FV1000 attached to an inverted microscope IX81 (Olympus, Tokyo, Japan, http://www.olympus.com). Images were obtained with Olympus UplanS-apo ×40 oil, 1.2 numerical aperture objective lens with sequential analysis to avoid crosstalk between image channels. Fluorophores were excited with a 488 nm line of argon laser (for Alexa fluor 488), 543 nm line of a HeNe laser (for Alexa fluor 568) and 633 line of a HeNe laser (for Topro-3). The emitted fluorescence was detected through spectral detection channels between 500 and 530 nm and 550 and 625 nm, for green and red fluorescence, respectively, and through a 650 nm long-pass filter for far red fluorescence.
Limiting dilutions and neurosphere-initiating cells assay
To determine the existence and the frequence of a “neurosphere-initiating cell” (NS-IC), we performed limiting dilution assays using single cell populations immediately after dissociation of neurospheres. Final cell dilutions ranged from 0.5 cell per well to 30 cells per well. Half amount of culture medium was replaced twice a week until day 21, after which the fraction of wells not containing neurosphere for each cell plating density was determined. These points were plotted against the number of cells plated per well. The number of cells required to form one neurosphere, which reflects the frequency of cancer stem cells in the entire population was then determined as described previously.33
Cells from GBM cultures were washed with serum-free DMEM/F12 medium and plated overnight with DMEM/F12 without supplement or growth factors. One fraction of these cells was incubated with EGF (50 ng/mL) for 30 min before fixation and permeabilization for immunostaining, or with bFGF (50 ng/mL), or a combination of both factors with the supplements N2 and B27. The unstimulated and stimulated cells were washed and fixed with PBS/paraformaldehyde (2%) at 37°C for 10 min and permeabilized by cooled methanol in ice for 30 min. After two washes in PBS/human serum albumin (0.5%), cells were stained with a first antibody (STAT3, pSer727-STAT3 and pTyr705-STAT3, 1:100, Cell Signaling) for 30 min at room temperature, then with a second FITC-labeled anti-mouse (BD Biosciences, San Diego, CA, http://www.bdbiosciences.com) or anti-rabbit IgG (Sigma). After washing in PBS/human serum albumin (0.5%), populations were analyzed by flow cytometry with a FACSCalibur (BD). The software used was CellQuest ProTM (BD). A total of 104 events were recorded for each sample. The background signal was determined using isotype control antibodies. Forward light scatter (FSC-height) of cells was monitored in association with their side light scatter (SSC-height) and fluorescence of STAT3 antibodies. For all samples, the amount of STAT3 or its phosphorylated forms was expressed as a percentage of the negative control.
Protein extracts were obtained from 106 cells by lysing samples in Laemmli buffer (Biorad, Marnes-La-Coquette, France, http://www.bio-rad.com). Protein samples were separated by SDS-PAGE and transferred onto a nitrocellulose membrane (Millipore, Billerica, MA, http://www.millipore.com). The membrane was blocked in Tris-buffered saline with 5% nonfat dry milk, and incubated with a primary antibody (STAT3, pSer727-STAT3 and pTyr705-STAT3, 1:250, Cell Signaling) for 2 hr at room temperature and horseradish peroxidase-conjugated secondary antibody (1:5,000) (Cell Signaling) for 45 min at room temperature. The anti-actin antibody (clone AC-15) was purchased from Sigma. Immunoreactive proteins were visualized using an enhanced chemiluminescence (ECL).
Magnetic cell sorting and flow cytometry of CD133-sorted cells
Cells were dissociated with accutase and resuspended in 1× PBS/0.5% BSA/2 mM EDTA. Magnetic labeling with 50 μL CD133/1-Biotin per 108 cells was performed using the Miltenyi Biotec Indirect CD133 Microbead Kit. Magnetic separation was carried out on the auto-MACS machine (Miltenyi Biotec). Aliquots of CD133+ and CD133− fractions were evaluated for purity by flow cytometry with 10 μL of CD133/2-phycoerythrin (Miltenyi Biotec) on the FACSCalibur (BD Biosciences).
Reverse transcription-quantitative polymerase chain reaction
Total RNA was extracted by TRIzol reagent (Invitrogen). First strand cDNA synthesis was carried out on 1 μg samples of total RNA using M-MLV reverse transcriptase according to the manufacturer's procedure (Promega, Madison, WI, http://www.promega.com). STAT3, BCL2, GFAP, CD133 or VEGF mRNA expression were normalized to GAPDH mRNA levels present in the same cDNA sample. BCL2, GFAP, VEGF, CD133 and GAPDH mRNA were determined by Taqman gene expression assays (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com). STAT3 primers were 5′-AGCAAAGAAGGAGGCGTCA-3′ (forward) and 5′ GCCGACAATACTTTCCGAATG-3′ (reverse). Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) was done in duplicate in a final volume of 25 μL using the Sybr Green PCR Master Mix or Taqman Master Mix (Applied Biosystems). The following cycling conditions were applied: an initial hold at 50°C for 2 min, followed by 10 min at 95°C and 40 cycles consisting of 95°C for 15 sec and 60°C for 1 min. Relative changes in STAT-3, VEGF, BCL2 and GFAP mRNA amounts were determined by the 2−ΔΔCt method.
Sense and antisense oligonucleotides from human Stat-3 coding region (+846, +866 from start codon) were annealed and introduced in pSuper plasmid, 3′ to polIII H1 promoter. polIII H1 promoter-shRNA Stat DNA fragment was then subcloned in the pTRIP/ΔU3-EF1α lentiviral vector encoding the green fluorescent protein (TRIP/ΔU3-EF1α-GFP). A shRNA directed against the luciferase protein was used as a control (Luc). Production of both shRNA-Stat and shRNA-Luc lentiviral vectors was performed as described.34, 35 CSC-GBM1 or CSC-GBM2 cells (5 × 105 per milliliter) were incubated with the indicated lentiviral particles at a multiplicity of infection (MOI) of 10 for 24 hr and then intensively washed. Two to three days following transduction, GFP+ cells were sorted by FACS with an ALTRA flow cytometer (Dako) and incubated for five additional days before assessing STAT3 expression by means of RT-qPCR or immunoblot analysis.
DNA was extracted from paired blood, tumor and neurospheres cells by using DNeasy blood or tissue kit (Qiagen). To analyze 9p21 locus and 10q23 locus, which are frequently deleted in gliomas, five microsatellite markers on 9p21 (D9S162, D9S1679, D9S1748, D9S157 and D9S171) and four microsatellite markers on 10q23 (D10S215, D10S541, AFMA and D10S1765) were used. DNA was amplified and aliquots of the PCR reactions were subjected to electrophoresis on a 3100 DNA Sequencer (ABI). The automatically collected data were analyzed with Genescan and Genotyper software (ABI). The LOH index was determined with the formula (height A2/A1 control/A2/A1 tumoral). Index <0.5 and >1.5 was considered as LOH.
- Top of page
- Material and Methods
- Supporting Information
In vitro and in vivo GBM-derived TISCs/CSCs analysis
Isolation and characterization of TISCs/CSCs were based on criteria described by Singh et al. and Galli et al.,5, 6 cells derived from tumors were cultured as proliferative nonadherent multicellular spheres in serum-free medium DMEM/F12 or NBE conditions (EGF and bFGF supplemented) for ∼1 week (Fig. 1a). These spheres expressed nestin and SOX2, but also CD133 at various levels (Fig. 1b) which was found to be highly expressed in primitive “stem” cell populations in patients with medulloblastomas and GBMs.4–6 After dissociation, these cells retain the capacity to proliferate and form secondary neurospheres. These cells were also able to differentiate into the three main cell lineages found in the central nervous system: neurons, astrocytes and oligodendrocytes using specific differentiation media as detailed in “Material and Methods” section (Fig. 1c). Cells from neurospheres were tumoral as revealed by May-Grümwald-Giemsa staining, showing many cytologic aberrations characteristics of tumor cells as they are grouped in clusters, with anisocytosis and anisocaryosis, an irregular nucleus, fine nucleoled-chromatin and many vacuoles in the cytoplasm. A lot of mitotic figures were noted consistent with the highly proliferative nature of these cells (Fig. 1e). These cells had the same genetic alterations than their respective paired tumor as CSC-GBM1 presents an LOH on chromosome 10q23 (marker D10S541) and CSC-GBM2 presents an LOH on chromosome 9p21 (marker D9S157) (Fig. 1d). Furthermore, tumorigenicity of the neurospheres was confirmed as nude mice developed tumors after orthotopic injection of 105 cells and 103 cells (Fig. 1f and Supporting Information Figure 1). Three mice were used for each experiment.
Quantification of neurosphere-initiating cells (NS-IC) in primary human GBMs
Doubling times for tumor-derived cells were 4.6 and 5 days for CSC-GBM1 and CSC-GBM2, respectively, and 4.5 days for U251MG (Fig. 2a). When cells forming a neurosphere were dissociated mechanically or enzymatically only a small percentage of them (reflecting the neural stem-like cells) could generate new neurospheres. To evaluate quantitatively the capacity to form novel neurospheres from dissociated neurospheres, we have designed a neurosphere formation assay using limiting dilution analysis. This assay was performed using the neurospheres obtained from both primary samples as well as those obtained from the U251MG cell line. We evaluated the frequency of neurosphere-initiating cells (NS-IC), corresponding to the number of cells required to generate at least one tumor sphere per well. This frequency was 1/9 for CSC-GBM1, 1/17 for CSC-GBM2 and 1/5 for U251MG (Table 1 and Fig. 2b). We determined whether the proportion of CD133-positive cells could influence the frequency of NS-IC. For that purpose, neurospheres were dissociated and analyzed by flow cytometry for CD133 expression before starting NS-IC assays. Our results suggest that the frequency of NS-IC is related to the percentage of CD133 positive cells. Indeed, the number of cells required to generate at least one sphere per well decreases when the percentage of CD133 positive cells increases (Fig. 2c).
STAT3 is constitutively phosphorylated in neurospheres
The expression of STAT3 and its phosphorylated forms were studied successively by flow cytometry, western blot and immunofluorescence. Cells were deprived of growth factors overnight and divided in two fractions, one fraction being stimulated with growth factors for 30 min before the analysis of STAT3 phosphorylation by flow cytometry and by western-blot analysis (Supporting Information Figure S2), and the unstimulated fraction used as a control.
Flow cytometry experiments indicated the presence of similar STAT3 expression in both tumor-derived and U251MG-derived neurospheres (Fig. 3a). The mean of individual results is represented as histograms in Figure 3b. Interestingly, the levels of Ser727-phosphorylated STAT3 were found surprisingly high, in the absence or presence of growth factor stimulation, with nearly 100% of cells labeled (Figs. 3a and 3b). As compared to Ser727-STAT3 phosphorylation, Tyr705-phosphorylated STAT3 levels remained relatively low in both patient-derived neurospheres but clearly detectable even in the absence of growth factor stimuli (Figs. 3a and 3b). These results were confirmed and extended by western-blot analysis. Total STAT3 protein expression in tumor-derived neurospheres and U251MG cell line was unaffected by a stimulating 30 min treatment for each growth factor analyzed, as compared to untreated cells, which have been deprived of growth factors by an overnight pretreatment. All these three cellular contexts exhibited strong constitutive Ser727 phosphorylation of STAT3, but much weaker amounts of phospho-Tyr705-STAT3. However, the levels of phospho-Ser727-STAT3 expression were enhanced on growth factor(s) stimulation in both patient samples as well as in U251MG cells (Supporting Information Figure S2). We further observed that phosphorylation of Ser727-STAT3 was not affected by the percentage of CD133-positive cells in neurospheres, and was always in the range of 80–100% of the cells labeled in flow cytometry assays (Fig. 3c). Confocal immunofluorescence experiments extended our previous observations and confirmed a high level of Ser727-phosphorylation of STAT3 in neurosphere-forming cells when using the phospho-Ser727-STAT3 antibody (Fig. 3d). Interestingly, Ser727-phosphorylation of STAT3 was more important in dividing cells (see magnification area in Fig. 3d). Phospho-Ser727-STAT3 was detected both in the nuclear as well as the cytoplasmic compartments.
Altogether these results show that in the absence of growth factors, STAT3 exhibits constitutive phosphorylation in TISC/CSCs isolated from GBMs, which preferentially affects its serine 727 residue. To further confirm these results, we performed flow cytometry analysis in three additional GBM stem cell lines (CSC-GBM3, CSC-GBM4 and CSC-GBM5) (Supporting Information Figure S3). All cell lines tested expressed elevated levels of STAT3 phosphorylation, suggesting that constitutive STAT3 activation is a common feature of CSC-GBM.
STAT3 inhibition strongly impairs proliferation and neurosphere initiation
We therefore examined the biological relevance of constitutive STAT3 phosphorylation in TISC/CSCs from GBMs by testing the influence of STAT3 inhibition on proliferation and neurosphere-initiating capacity of these cells. For that purpose two strategies were used, we first inhibited STAT3 activity using a small chemical compound, Stattic, which acts by blocking the SH2 domain of STAT3, inhibiting its phosphorylation, its dimerization and nuclear translocation.36 We assessed the inhibition of phosphoSTAT3 by Stattic by performing western-blot analysis (Supporting Information Figure S4). After 5 days of treatment with Stattic at 5 μM as well as at IC50 concentrations (see below), phosphorylation levels of serine 727 and tyrosine 705 residues decreased significantly in both CSC-GBM1 and 2. Treatment of these cells with Stattic correlates with a reduced expression of VEGF, which is known to be a target gene of STAT3, confirming the efficiency of the drug at inhibiting STAT3 activation (Supporting Information Figure S5A). In addition, we determined the IC50 of Stattic in both patient-derived neurospheres CSC-GBM1 and CSC-GBM2. After 5 days of treatment, which corresponds to the doubling time of these cells, only 50% of cells were viable using Stattic at concentrations of 1.08 and 2.5 μM, respectively. Moreover, at these concentrations we observed an inhibition of proliferation as soon as 48 hr after the beginning of the treatment (Fig. 4a). We further tested the consequences of Stattic-mediated STAT3 inhibition on the functionality of neurosphere-initiating cells. The ability of TISC/CSCs to form neurospheres decreased and was eventually abrogated since no neurosphere was observed after 21 days of treatment with either CSC-GBM1 or CSC-GBM2 (Fig. 4b and Table 1). To analyze the mechanisms of cell death we performed annexin staining of CSC cells after treatment with Stattic. After 2 or 5 days of treatment with Stattic CSC-GBM1 and CSC-GBM2 cells undergo apoptosis at various levels. As shown in Supporting Information Figure S5B, similar basal levels of apoptosis are present in both CSC-GBM in the absence of Stattic, however, after 5 days of treatment CSC-GBM1 exhibit 10.7% of apoptotic cells versus 54.8% in CSC-GBM2. This observation is consistent with the inhibition of neurosphere formation following Stattic treatment as this inhibition is more prominent in CSC-GBM2. To confirm these results, we aimed at inhibiting all STAT3 biological activities by suppressing STAT3 expression. Lentiviral vectors encoding for STAT3-directed shRNA were used to transduce patient-derived neurospheres CSC-GBM1 and CSC-GBM2. These vectors are known to induce a massive suppression of STAT3 protein expression in primary human CD34+ hematopoietic progenitors, with only partial inhibition of STAT3 transcripts (data not shown). The inhibition of STAT3 expression was controlled by RT-qPCR for CSC-GBM1 and western-blot for CSC-GBM2. We observed a total inhibition of STAT3 expression in CSC-GBM2 cells and a 50% decrease of STAT3 transcripts in CSC-GBM1 cells, as compared with untransduced cells (Fig. 5a). We then tested the effect of shRNA-STAT3 on proliferation and on neurosphere forming capacity of CSC-GBM1 and CSC-GBM2. We observed an inhibition of proliferation of the neurospheres transduced with the anti-STAT3 shRNA encoding vector while those transduced with the control anti-Luciferase shRNA encoding vector proliferated normally, as compared to untransduced cells (Fig. 5b). Similarly, the number of cells required to generate at least one tumor sphere per well strongly increased in cells transduced with shRNA-STAT3 whereas in cells transduced with shRNA-Luc this frequency was comparable to that of untransduced cells (Fig. 5c and Table 1). Altogether, these observations indicate that suppressing STAT3 strongly affects the functionality of NS-IC from primary human GBMs.
Combination of anti-STAT3 Stattic and TMZ induces enhanced inhibitory effects on proliferation of brain tumor stem cells
TMZ is the most commonly used chemotherapeutic agent in the therapy of GBMs.1 TMZ is an alkylating agent that achieves its cytotoxic effect mainly by methylating the O6 guanine position. Resistance to this treatment was shown to be due to the methylation of DNA repair gene MGMT (O6-methyl-guanine-methyltransferase).37–39 Moreover, GBM stem cells and progenitor cells are highly resistant to conventional chemotherapeutic drugs, including TMZ.12–14 In this regard, we asked whether the combination of Stattic with TMZ affects the proliferation rate of patient-derived neurospheres. After IC50 determination for TMZ in patient-derived neurospheres CSC-GMB1 and CSC-GMB2 (612 μM and 537 μM, respectively), we used lower concentrations (IC20 concentrations) of Stattic and TMZ to evaluate possible synergistic or additive effects. At IC20 concentrations, when Stattic or TMZ were added separately, 80–100% of cells were viable. Interestingly, at these concentrations when both drugs were combined we observed a much stronger decrease of the proliferative rates of CSC-GMB1 and CSC-GMB2, which dropped to 50% in comparison with untreated cells. This suggests an additive effect of these drugs. Stattic may thus sensitize CSC-GBM cells to the action of TMZ (Fig. 6a). Moreover, when shRNA-STAT3 transduced CSC-GBM2 were treated with IC20 concentrations of TMZ the proliferative rate of the cells similarly dropped to 57% in comparison with untreated cells, whereas 96% of cells transduced with shRNA-Luc and treated with the same concentrations of TMZ were viable (Fig. 6b). Thus inhibition of STAT3 sensitizes brain tumor stem cells to the action of TMZ. We confirmed these results by performing clonogenicity assays on CSC-GBM1 cells in methylcellulose medium. As shown in Supporting Information Figure S6, a clonogenic index of 79% was observed after treatment with IC20 concentration of Stattic, consistent with results shown in Figure 6. In accordance with the cytostatic properties of TMZ, the clonogenicity index found after treatment with the concentration previously determined as IC20 in liquid medium is much below 80%. Nevertheless, accordingly to previous results, after treatment with Stattic and TMZ at IC20, we observed the additional effect of both drugs as the clonogenic index dropped to 2.5%. In addition, we measured the proportion of CD133+ cells before and after treatment by QRT-PCR. We observed a slight decrease of CD133 mRNA expressing cells when treated with IC20 concentrations of Stattic or TMZ. This decrease is statistically significant when Stattic and TMZ are added together, suggesting that this association of drugs depletes more efficiently CD133 expressing stem cells (Fig. 6c).
- Top of page
- Material and Methods
- Supporting Information
GBMs have been shown to harbor some primitive tumor initiating or “neural-stem-like” cells which are able to give rise to clonal populations on serial transplantations in immune deficient/compromised mice suggesting their self-renewal capacity.4–8 First reports suggested that this capacity was confined to the CD133+ cells present within GBMs.4–6 However, recent studies showed that CD133− cells from gliomas are also tumorigenic.10, 11 Neurospheres have been shown to be able to grow in clonal conditions40, 41 but after sequential passages it is difficult to determine if a clonal dominance phenomenon occurs. Nevertheless, in this work, we have shown, with reproducible data obtained from three independent experiments and with different patients, that “neurosphere-initiating cells” (reflecting TISC/CSCs) are present with a frequency of 1/9 to 1/17. Such a frequency may explain why GBM are so refractory to all therapeutic treatments, either surgery, RT or CT. This in vitro assay and the quantification of NS-IC can be used as a surrogate assay to determine the sensitivity of these cells to potential anti-GBM drugs, which could be of major clinical application in the future. The “cancer stem cell” hypothesis has not been validated entirely in all tumors and the normal tissue hierarchy cannot be entirely found in neoplastic counterparts of all tumors.42 GBM could be one of the candidates in favor of cancer stem cell hypothesis. Further work should aim at deciphering the molecular events involved in the self-renewal of this cell population.
Our data further indicate that independently of their CD133 status, neurospheres derived from GBM express a constitutively phosphorylated STAT3 harboring transcriptional activity as indicated by STAT3-dependent VEGF expression. More interestingly, using two different methods we provide evidence that STAT3 is essential for the proliferative and the self-renewal capacity of the neural stem cells able to form neurospheres since: (i) a chemical compound, Stattic, that inhibits STAT3 activation by blocking its essential SH2 domain, and (ii) introduction of lentiviral vectors expressing STAT3-directed shRNA, that inhibits STAT3 expression, both inhibit the functionality of NS-IC. Several studies have shown that blockage of STAT3 signaling pathway by either chemical inhibitor, dominant-negative mutant protein of STAT3 or transcription factor decoy oligodeoxynucleotides suppresses growth of malignant glioma cells.31, 32, 43–46 Although this paper was submitted, Sherry et al.47 reported that STAT3 is required for proliferation and maintenance of multipotency in GBM stem cells. Thus STAT3 activation appears as a common feature of CSC-GBM.
These authors reported that STAT3 inhibition in GBM-derived stem cells with STA-21 or S3I-201 causes no increase in apoptosis. In contrast to these results, we observed various levels of apoptosis in CSC-GBM1 and CSC-GBM2 cells after 2 or 5 days of treatment with Stattic. These contradictory results could be explained by the different intrinsic activities of these compounds, although Stattic, STA-21 and S3I-201 are targeting the same SH2 domain of STAT3 the last two compounds have additional properties as they contain polyphenol groups. Polyphenols have been shown to act as free radicals scavengers and several reports demonstrated their crucial role in preventing apoptosis in mammalian cells.48
Together these results suggest that strategies aimed at inhibiting STAT3 could have important practical implications in the treatment of this malignant tumor. Others have reported that an oligonucleotide which includes a selective responsive element for STAT3-dependent promoters can also work as an inhibitory bait for STAT3, and thus acts as an efficient inhibitor of tumor growth in vitro, as well as in vivo when injected within the tumor.31, 32 Research efforts on stabilizing such nucleic-based compound in biological fluids and on increasing their entry within target cells may soon provide additional helpful antitumor chemicals.
In addition, our data show that inhibition of STAT3 has an additive effect with TMZ to inhibit NS-IC. Moreover, the proportion of CD133+ cells decreased after treatment with Stattic and TMZ indicating that this population can be more efficiently targeted by the association of both drugs than TMZ alone, which is the commonly used chemotherapeutic agent in the therapy of GBMs.
Although TMZ is the most commonly used chemotherapeutic agent in the therapy of GBMs, our data indicate that inhibition of STAT3 has an additive effect with TMZ to inhibit NS-IC by depleting more efficiently CD133 expressing stem cells.
The mechanism by which STAT3 maintains NS-IC and is acting in GBM cells remains to be determined. STAT3 is commonly activated by a large range of hematopoietic (IL-3, GM-CSF, TPO, etc.) and nonhematopoietic cytokines and growth factors (Prolactin, EGF, LIF, etc.), on binding to their respective transmembrane receptors. STAT family members are tyrosine and serine phosphorylated, undergo selective dimerization and translocate to the nucleus where they bind to consensus sequences on the promoters of target genes providing transcriptional regulation. Expression and activity of some of these STATs, such as STAT3 or STAT5a/b, have been shown to be strongly increased in a variety of malignancies. They also participate to the oncogenic process, and oncogenic STAT3 and STAT5a/b have been shown to be important direct regulators of expressions of cell survival factors (BCL-x, surviving, etc.) and actors involved in cell proliferation (c-Myc, cyclin D1, etc.).21, 22, 49–51 STAT3 has also been reported to act through noncanonical signaling pathways, involving indirect interactions with transcription factors such as NFkB,23, 52 and recently cytoplasmic activities of STAT3 have been shown through association with the mitochondrial respiratory chain components or other signaling molecules.53, 54 We observed that in NS-IC STAT3 is mainly phosphorylated on Ser727 rather than Tyr705. It is known that the phosphorylation of STAT1 or STAT3 on Ser-727 in the presence of Tyr phosphorylation of these STATs leads to an enhancement of their transcriptional activity over that seen with the tyrosine-phosphorylated STATs alone.55, 56 Alternatively, the Ser phosphorylation of STAT3 has been recently shown to have another function independent of its role in modulating transcriptional activation. This phosphorylation is crucial to the function of mitochondrial STAT3 in cellular respiration, by augmenting electron transport chain activity. It moreover sustains altered oxidative phosphorylation activities characteristic of cancer cells and supports RAS-dependent malignant transformation.53, 57 This latter activity is largely dependent on Ser-727 phosphorylation but independent of Tyr phosphorylation of STAT3.
Moreover, we have shown, using confocal microscopy, that P-Ser STAT3 was present in both the nuclear and cytoplasmic compartments of GBM cells, in accordance with recent observations in leukemic samples.58 Future investigations will help to determine whether the activation of STAT3 is a late or an early event in the pathogenesis of GBM, and to unveil the molecular mechanisms by which STAT3 synergizes with TMZ. Collectively, our work reveals that STAT3 represents a promising molecular target for suppressing GBM-initiating tumor stem cells. Such a knowledge could be of use to limit relapse rates of GBMs and eventually to extend the median survival.
- Top of page
- Material and Methods
- Supporting Information
We thank Pr. F. Berger and team for their invaluable help with in vivo tumorigenicity assays. Ms. A. Cantereau (UMR CNRS 6187) for her assistance with confocal microscopy. Mr. S. Martin, Mr. P. Rivet and Ms. C. Marquant for technical assistance with cell culture.
- Top of page
- Material and Methods
- Supporting Information
- Top of page
- Material and Methods
- Supporting Information
Additional Supporting Information may be found in the online version of this article.
|IJC_25416_sm_suppinfofig1.tif||125K||Supporting Figure 1.|
|IJC_25416_sm_suppinfofig2.tif||199K||Supporting Figure 2.|
|IJC_25416_sm_suppinfofig3.tif||341K||Supporting Figure 3.|
|IJC_25416_sm_suppinfofig4.tif||150K||Supporting Figure 4.|
|IJC_25416_sm_suppinfofig5a.tif||42K||Supporting Figure 5a.|
|IJC_25416_sm_suppinfofig5b.tif||212K||Supporting Figure 5b.|
|IJC_25416_sm_suppinfofig6.tif||103K||Supporting Figure 6.|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.