Telomestatin Impairs Glioma Stem Cell Survival and Growth through the Disruption of Telomeric G-Quadruplex and Inhibition of the Proto-oncogene, c-Myb

This article is featured in Highlights of This Issue, p. 1187

The development of effective therapies for glioblastoma multiforme (GBM) is a challenging endeavor due to the aggressive proliferation and the high migratory potential of this form of cancer. Recent studies have suggested the existence of a hierarchical organization of multiple heterogeneous cell populations in GBMs having distinct tumor-driving capacities (1). Among heterogeneous tumor cells, glioma stem cells (GSC) are defined as a subpopulation that is capable of self-renewal and differentiation into multilineaged tumor cells with distinct tumorigenic potentials in vivo. The resistance of GSCs to current forms of therapy is likely related to the failure of current treatments (2). Identification of novel therapeutic strategies for GSCs in GBMs remains a major hurdle to effectively attack this highly malignant tumor. GSCs exhibit phenotypic and genetic similarities to their somatic counterparts, neural stem cells (NSC). Targeting shared pathways that regulate the survival of both GSCs and NSCs, therefore, may eradicate both types of stem cells. To accomplish the selective targeting of GSCs over NSCs, it is crucial to uncover the mechanisms that specifically regulate the initiation of GBMs and the maintenance of the stem cell–like phenotype of GSCs.

Human chromosomes are capped with telomeres, regions of repeating DNA that prevent the degradation of genes at the chromosomal ends (3). Because of incomplete replication of linear DNA ends, human somatic cells progressively shorten their telomeres during every round of cell division. Immortalized cancer cells have an increased ability to extend their telomeres, which is one of the reasons that cancer cells can bypass replication senescence and subsequent cell death (4, 5). If indefinite cell division is a property that is unique to tumor stem cells, one potential approach to target GSCs would be to disrupt telomere extension. In the process of telomere elongation, the enzyme telomerase plays a critical role. Despite the emerging hope for attacking telomerase in cancers based on this unique character, the major limitation of this approach in clinical settings is its delayed effectiveness.

Telomestatin is a natural product isolated from Streptomyces anulatus 3533-SV4 (6) and stabilizes the G-quadruplex (7) that is postulated to be present in telomeric DNA (3) and in the promoter regions of several proto-oncogenes (8–11). Formed G-quadruplex structures function as transcriptional repressor elements (12). Treatment with telomestatin induces apoptosis of various cancer cells with relatively less of an effect on somatic cells (13, 14). Although the effect of telomestatin on telomeric DNA has been well described, it is not clear whether it is the only mechanism of higher sensitivity of cancer cells over somatic cells. In addition, the sensitivity of cancer stem cells to telomestatin has not been shown yet.

Here, we show that telomestatin triggers the preferential apoptosis of GSCs with less of an effect on normal precursors or non-GSCs in GBMs. Immunofluorescence in situ hybridization (iFISH) detected the presence of damage in both telomeric and non-telomeric DNA regions in GSCs but not in non-GSCs. Analysis of a cDNA microarray identified a reduction in the proto-oncogene, c-Myb, following telomestatin treatment of GSCs. Decreased c-Myb expression was also observed in pharmacodynamic analyses of telomestatin-treated xenografted tumors. Moreover, treatment of tumor-bearing mice showed a statistically significant reduction in tumor sizes in vivo. Finally, immunohistochemistry of clinical samples of GBMs and normal tissues exhibited a statistically significant elevation of c-Myb levels in tumors. Collectively, these data suggest that targeting GSCs with telomestatin has the potent effect on diminishing GBM growth in vitro and in vivo through disruption of telomeric DNA and inhibition of c-Myb.

Characteristics of 8 GBMs (GBM146, 157, 205, 206, 218, 1600, 2313, and 13) and 2 non-tumor human fetal brain specimens (f16w and 1105A) from aborted fetus were published previously (15–18). Methods for establishment of GBM sphere cultures and normal sphere cultures are published previously (16, 18) and details are described in Supplementary Information. All the experiments were carried out with short-term cultures within 20 passages.

Telomestatin was synthesized in the Biomedicinal Information Research Center, National Institute of Advanced Industrial Science and Technology (Tokyo, Japan). The quality of telomestatin was assessed with HPLC (UPLC) system and the readout was based on the UV absorptions. The structure of telomestatin was also confirmed by ¹H and ¹³C NMR spectra. Other anticancer drugs and chemicals were purchased from Sigma-Aldrich and dissolved in dimethyl sulfoxide (DMSO).

Detailed methods are published previously (16, 18–20) and described in Supplementary Information.

Brain slice invasion assay was conducted as described previously (21). Briefly, GBM spheres of 300 to 400 μm diameter were treated with telomestatin or vehicle (DMSO) for 24 hours. Coronal brain slices obtained from neonatal mice were prepared for organotypic culture. Viable spheres that did not stain with propidium iodide were washed in slice medium, manually placed on the brain slices, and imaged by fluorescence microscopy every 24 hours. Cells from spheres treated with telomestatin were also assessed for cell viability using an assay for reduction of soluble tetrazolium (Cell Titer kit, Promega).

RNA was extracted from GBM sphere samples treated with DMSO or 1 μmol/L telomestatin for 24 hours with RNeasy Mini Kit following the manufacturer's protocol (QIAGEN). RNA samples were analyzed by the FGC using an Agilent 2100 Bioanalyzer Lab-On-A-Chip Agilent 6000 Series II chip to determine the integrity of the samples. Sample labeling and hybridization were conducted according to the manufacturer's protocols.

Microarray slides were hybridized overnight, washed, and then scanned with Agilent G2505C Microarray Scanner. The information about each probe on the array was extracted from the image data using Agilent Feature Extraction 10.5 (FE). The raw intensity values from these files are imported into the mathematical software package “R,” which is used for all data input, diagnostic plots, normalization, and quality checking steps of the analysis process.

The slides of GBM tissue microarray were obtained from Department of Pathology at the Ohio State University (OSU; Columbus, OH). After deparaffinization by xylene, immunohistochemistry with the c-Myb antibody was carried out as described previously (22). Isogenic IgG control was used to confirm the specific signals for c-Myb immunoreactivity. The staining intensities were blindly categorized into 3 groups (negative, single positive, double positive) by 2 independent neuropathologists.

Values are given as means ± SD, unless noted otherwise in the figure legend. The number of replicates is noted in the figure or legends. Absent error bars in the bar graphs signify that SD values are smaller than the graphic symbols. Comparison of mean values between multiple groups was evaluated by a Tukey honestly significant difference (HSD) test for post-ANOVA pairwise comparisons in a one-way ANOVA unless noted otherwise in the figure legend. Comparison of mean values between 2 groups was evaluated by χ² test or Student t test. All statistical tests were two-sided. For all statistical methods, a P value less than 0.05 was considered significant.

The effect of telomestatin in vitro was first tested on a panel of 39 human cancer cell lines (JFCR39; Fig. 1A; ref. 23). Cells derived from tumors of the central nervous system (CNS) exhibited higher sensitivity than others. With these brain tumor cell lines, the concentration of telomestatin required for a 50% growth inhibition (GI₅₀; concentration needed to reduce the growth of treated cells to half that of untreated cells) ranged between 1 and 10 μmol/L (Supplementary Fig. S1A). With patient-derived short-term GBM cell cultures (GBM1600 and 2313) propagated in serum-containing medium, the GI₅₀ value was approximately 5 μmol/L (Supplementary Fig. S1B). This result suggests that telomestatin is a relatively potent and selective inhibitor of brain tumor–derived cells compared with other cancer-derived cells. Next, the sensitivity of GSCs to telomestatin was examined. Sphere-forming potential is one unique characteristic of GSCs (24, 25). Using short-term cultures derived from specimens of 5 patients with GBMs, we investigated the effect of varying doses of telomestatin on sphere formation (Supplementary Fig. S1C). Remarkably, in all samples, treatment with 1 μmol/L telomestatin completely abolished sphere formation. In 3 of these 5 samples, treatment with 0.1 μmol/L of telomestatin significantly reduced sphere numbers as well (P < 0.05; Supplementary Fig. S1C). These results suggest a prominent inhibitory effect of telomestatin on GSC phenotypic sphere formation capability in vitro.

In several organs, a differential sensitivity to telomestatin treatment was previously observed between tumor cells and their somatic counterparts (13, 14). We therefore sought to compare the effects of telomestatin treatment on the growth of patient-derived GSCs and normal neural precursors derived from human fetal brains. Normal neural precursors and GSCs were enriched in serum-free media with the supplement of basic fibroblast growth factor and epidermal growth factor (Supplementary Fig. S2). When both of these cultures were stimulated with 10% FBS for differentiation, multiple types of lineage-committed cells were observed, as seen by a decline in Sox2 expression and upregulation of neuron-specific class III β-tubulin (TuJ1) and glial fibrillary acid protein (GFAP; Supplementary Fig. S2). Following examination of these precursor cells, we found that compared with GSCs, neural precursor cells were relatively resistant to telomestatin treatment (Fig. 1B). The graph in Fig. 1B shows a significant difference of relative cell numbers between 3 GSC samples (GBM146, 157, and 206) and 2 normal neural precursor samples (f16w and 1105A) after treatment with 1 μmol/L telomestatin for 96 hours. These data suggest that telomestatin treatment strongly impairs GSC growth, whereas normal neural precursors are less sensitive to telomestatin treatment. Likewise, we verified the effective dose of telomestatin required to affect the growth properties of human astrocyte primary cultures and immortalized human astrocytes overexpressing E6, E7, and hTERT (normal human astrocytes; NHA). One phenotypic difference between these two cultures is that only NHA cells are tumorigenic in immunocompromised mouse brains (26). Following incubation of these 2 cultures with varying doses of telomestatin for 96 hours, we found that normal astrocytes are more resistant to treatment than NHA cultures (Supplementary Fig. S1D).

A cell surface protein, CD133, is not a universal marker for GSCs in GBMs (27); however, we have found that in some of our clinical samples, positive CD133 expression correlates well with properties of GSCs, including tumorigenicity, in vitro clonogenic potential, multilineage differentiation capability, and high expression of stem cell–related genes and proteins (15, 18). We therefore reasoned that if telomestatin preferentially eradicates GSCs, the proportion of CD133-positive cells among total GBM cells should decrease with treatment. To explore this possibility, we treated tumor spheres derived from the 3 GBM cases with varying doses of telomestatin for 48 hours and conducted fluorescence-activated cell sorting (FACS) to determine the proportion of CD133-positive cells. As expected, telomestatin treatment reduced the proportion of CD133-positive cells in a dose-dependent manner with the proportion of CD133-positive cells being approximately half of the control samples at 1 μmol/L concentration (91.5% vs. 62.2% in GBM146, 92.6% vs.43.5% in GBM157, and 60.7% vs. 24.3% in GBM206; Fig. 1C, top, and Supplementary Fig. S3A). By separation of GBM157 cells based on cell surface CD133 expression, we confirmed that sphere-forming potential is restricted to CD133-positive cells, but not CD133-negative, cells (Supplementary Fig. S3B). When treated with telomestatin (1 μmol/L), these sphere-forming cells were eliminated (Supplementary Fig. S3B). Immunocytochemistry with a CD133 antibody and Hoechst dye showed that telomestatin treatment resulted in the condensed fragmentation of nuclei of CD133-positive cells, suggesting the induction of apoptosis, but not differentiation of CD133-positive cells into CD133-negative cells by telomestatin (Fig. 1C, bottom). Taken together, these results show that telomestatin treatment preferentially eradicates GSCs through an apoptotic mechanism.

We then sought to determine the effect of telomestatin on GSC self-renewal. One means of estimating in vitro self-renewal is through the use of serial assays to test multiple rounds of sphere-forming capacity (28). We analyzed the capacity of these cells to form secondary spheres after telomestatin treatment (Fig. 1D). Following 7 days of incubation with telomestatin, we carried out secondary sphere-forming assays using drug-free culture conditions. A reduction in secondary sphere numbers was observed with even less concentrations of telomestatin (0.2–0.5 μmol/L), indicating that treatment with lower doses of telomestatin abolishes self-renewal capacity of GSCs in vitro.

We next sought to test the effect of telomestatin treatment on tumor cell growth in vivo. First, we conducted transplantation of 2 GBM sphere samples (GBM146 and 157) pretreated with telomestatin (1 or 5 μmol/L) into immunocompromised mouse brains in accordance with Institutional Animal Care and Use Committee–approved protocols at OSU (Supplementary Fig. S4). These tumor-bearing mice were then sacrificed at 3 months posttransplantation, and sizes of resultant tumors were determined by immunohistochemistry with human-specific nestin and vimentin antibodies. The average size of intracranial tumors was reduced with the lower dose (1 μmol/L; Supplementary Fig. S4B and S4E) and almost completely abolished with the higher dose (5 μmol/L) of telomestatin (Supplementary Fig. S4C and S4F). Similar results were observed with the GBM146-derived model treated with 5 μmol/L telomestatin (Supplementary Fig. S4D and S4G). Collectively, pretreatment of GSCs with telomestatin potently inhibits their in vivo tumor formation and growth.

Second, we evaluated the therapeutic efficacy of telomestatin by injection into the intracranial tumor cavities of immunocompromised mice. The mouse intracranial tumors were created by implantation of GBM157 spheres into the striatum (Fig. 2A). We then injected telomestatin at 2 different time points; at day 0 (on the day of tumor implantation) and at day 14. The rationale for choosing these 2 time points is to investigate separately the effect of telomestatin on both tumor initiation and growth of established tumors. We then measured tumor sizes 3 months after transplantation. The effect of the treatment was evaluated by immunohistochemistry using a human-specific nestin antibody and a vimentin antibody (Fig. 2B; ref. 18). In both cases, local injection of telomestatin resulted in a 55% to 71% reduction in the overall tumor size, determined by nestin immunoreactivity (Fig. 2B, top). Similarly, a 72% to 76% reduction was observed with vimentin staining (Fig. 2B, bottom). To verify whether this procedure produces any cytotoxic effects on normal brain, we conducted immunohistochemistry for activated caspase-3, a marker for apoptosis, in mouse brains bearing tumor cells 2 days after telomestatin injection. Activated caspase-3 immunoreactivity was strictly restricted within the telomestatin-injected tumor lesions and not in adjacent normal brain tissues including the subventricular zone of the lateral ventricles (Fig. 2C, top). These data were in concurrence with the results of Ki-67 staining which shows a reduction in proliferating cells in telomestatin-treated tumors (Fig. 2C, bottom left). In control tumors treated with DMSO, few cells were positive for activated caspase-3, even in the area of tumor cell injection (Fig. 2C, bottom right). In summary, we confirmed that intratumoral telomestatin treatment significantly diminishes tumor size in vivo without causing the apparent death of normal cells.

One of the hallmarks of human GBMs is its invasive and infiltrative nature. In agreement with recent studies that showed an enhanced ability of GSCs for migration and invasion (29), GSC-derived tumor models exhibit aggressive migration upon transplantation into mouse brains (Fig. 3A). Of note, telomestatin-treated tumor cells showed significantly less migration into the contralateral hemisphere (Fig. 3A, bottom). This could be due to a reduced number of live tumor cells and/or reduced invasive capacity. To further study this observation, we conducted an assay for cell invasion on organotypic cultures of brain tissue which mimics brain cytoarchitecture and natural barriers to cell movement (Fig. 3B; ref. 21). First, we confirmed that undifferentiated GBM spheres (GSC) possess greater migration potential into mouse brains than non-GSCs from the same samples (Fig. 3C). We then treated GBM146 and 157 spheres with 3 different doses of telomestatin for 24 hours, seeded them onto mouse brain slices, and used fluorescence microscopy to evaluate invasion for up to 72 hours, as previously described (30). Both GBM sphere samples dispersed aggressively through brain tissue in the absence of the drug (Fig. 3D). In contrast, telomestatin significantly reduced cell migration in a concentration-dependent manner. Assessment of viability by in situ fluorescence and parallel viability assays indicated that this effect was specific for cell migration; as telomestatin did not affect viability of nondissociated spheres until 48 hours (5 μmol/L) to 96 hours (1 μmol/L) after treatment (data not shown). Collectively, these data indicate that telomestatin abrogates both growth and migration of GSCs in vivo and in vitro.

Although previous studies have reported several different mechanisms of action for telomestatin, it appears to kill non-CNS tumor cells primarily through DNA damage–derived apoptosis (13, 14, 31–33). When GBM sphere samples (GBM146, 157, and 206) were treated with telomestatin, we found a 17% to 35% increase in the number of cells with fragmented nuclei by Hoechst dye staining, a typical morphologic feature of apoptotic cell death (Fig. 4A). An abundance of apoptotic cells in telomestatin-treated GBM146 and 157 spheres was also confirmed by staining with fluorescein-conjugated Annexin V antibody and labeling with propidium iodide (Supplementary Fig. S5A). Western blot analysis showed an induction of p53 in telomestatin-treated GBM spheres (Supplementary Fig. S5B).

Because our data in Fig. 1 suggest that the potency of telomestatin is predominant in CNS tumor–derived cell lines and GBM spheres, we hypothesized that telomestatin has previously uncharacterized molecular targets in the non-telomeric DNA of GBM cells. To address this question, we conducted the iFISH assay that enables the simultaneous localization of telomeres and 53BP1, a marker of the DNA damage response (34). As shown in Fig. 4B, DMSO-treated control GBM146 GSCs exhibited uniform distribution of 53BP1 in the nucleoplasm, indicating that these cells did not initiate the DNA damage response under normal growth conditions. Conversely, telomestatin-treated GSCs rapidly developed punctate, nuclear 53BP1 foci in a dose-dependent manner (Fig. 4B). Importantly, some of these foci colocalized with non-telomeric DNA (Fig. 4C), therefore representing both telomeric and non-telomeric dysfunction-induced foci, a hallmark of deprotected DNA damage (35, 36). Meanwhile, consistent with their lower susceptibility to telomestatin-induced apoptosis, non-GSCs in GBM146 did not exhibit the 53BP1 foci upon treatment with telomestatin or DMSO under the identical culture condition (Fig. 3B and data not shown). Taken together, these results indicate that telomestatin induces telomeric and non-telomeric DNA damage preferentially in GSCs, and the loss of tumor stemness is likely associated with a failure in the DNA damage response in GSCs elicited by telomestatin.

The presence of 53BP1 foci in DNA regions separate from telomeres led us to search for target genes of telomestatin. To this end, we carried out cDNA microarray experiments using 4 GBM sphere samples with or without telomestatin treatment. Specifically, we focused on changes in G-quadruplex–interacting genes and telomere-related genes. Among the 15 genes identified, only c-Myb expression was significantly decreased in all 4 telomestatin-treated GBM sphere samples (Fig. 5A). An analysis of genes that were previously reported as stemness-related genes or proneuronal, mesenchymal, or proliferative GBMs (36) did not identify any gene whose expression was significantly decreased by telomestatin treatment (Fig. 5A). Downregulation of c-Myb in GBM spheres was further validated by reverse transcriptase PCR (RT-PCR) and Western blot analysis in all 4 samples (Fig. 5B and Supplementary Fig. S6). Conversely, c-Myb expression was not affected in telomestatin-treated non-GSCs or normal precursors, and as expected, the basal levels of c-Myb expression in these cell populations were substantially lower than those in GSCs (Fig. 5B). Immunocytochemistry showed the presence of c-Myb expressing cells in nestin-positive GBM157 spheres (Fig. 5C, left) and the decline in c-Myb expression in telomestatin-treated GBM spheres was also confirmed by immunocytochemistry (Fig. 5C, right). In agreement with decreased expression in vitro, pharmacodynamic evaluation of mouse intracranial tumors at day 2 following telomestatin injection showed an in vivo reduction of c-Myb immunoreactivity (Fig. 5D). These data support the idea that c-Myb is another molecular target of telomestatin and suggest that c-Myb is involved in survival and/or growth of GSCs in GBMs.

We then sought to determine the functional role of c-Myb in GSCs both in vitro and in vivo. First, we evaluated the effect of c-Myb overexpression and knockdown with GBM157 spheres in vitro. c-Myb knockdown resulted in abrogated growth of GBM157 sphere cells, which phenocopied the results with telomestatin treatment (Fig. 6A, left, and Supplementary Fig. S7). In turn, c-Myb overexpression in telomestatin-treated GBM157 spheres resulted in a full recovery of sphere cell growth at 0.5 μmol/L of telomestatin and a partial recovery at 1 μmol/L of telomestatin (Fig. 6A, right, and Supplementary Fig. S7). Next, we evaluated the effect of c-Myb knockdown on in vivo tumor growth. To this end, another short-term GBM sphere culture, GBM13, was infected with lentivirus harboring short hairpin RNA for either c-Myb or a nontargeting sequence (Supplementary Fig. S7; ref. 26), and the survival periods of tumor-burdened mice were determined (Fig. 6B). Consequently, lentivirus-derived c-Myb knockdown prolonged the survival of mice with GSC-derived intracranial tumors in a statistically significant manner. Taken together, these data suggest that c-Myb plays an essential role in GSCs growth in vitro and in vivo and is a target of telomestatin in GSCs.

To evaluate the clinical relevance of the function of c-Myb in GBMs, we investigated the expression of c-Myb in 66 surgical specimens of GBMs and 24 normal brain tissues (Fig. 6C). We conducted immunohistochemical staining with these samples using an antibody for c-Myb (22). c-Myb immunoreactivity was specifically observed in the nuclear compartments in both samples. Representative pictures display the typical pattern of the staining (Fig. 6C). As a result, c-Myb was abundantly expressed in nuclei of GBM cells in 62% of samples, whereas only 25% of normal brains showed positivity of c-Myb immunoreactivity (Fig. 6D; P = 0.004).

The cancer stem cell hypothesis postulates the existence of a subpopulation of poorly differentiated tumor cells with the capacity to self-renew and repopulate a diversity of lineage-committed tumor cells. By definition, in vivo tumor formation in immunocompromised mouse brains is the only direct evidence for the presence of GSCs within a pool of heterogeneous tumor cells. However, it is possible that this assay may select for tumor cells that have escaped from the immune system and/or human cells that have adapted to the microenvironment in mouse brains. These issues aside, the therapeutic significance of telomestatin treatment for human GBMs is suggested by its effect on in vivo growth in mouse brains (Fig. 2). To obtain clinically relevant data concerning the effect of telomestatin treatment, we conducted intratumoral injection of telomestatin. As a result, tumor size was significantly decreased. However, this treatment did not completely eliminate the tumors. It is possible that GSCs may contain a subpopulation of telomestatin-resistant cells. The relatively small number of samples tested in this study has allowed only limited interpretation concerning the genetic and phenotypic differences of GSCs as well as how universally effective telomestatin treatment would be on human patients with GBMs. To determine the therapeutic window of telomestatin, we need to address the question of whether or not the normal brain can tolerate doses of telomestatin required to kill tumor cells.

The cell-sorting technique using cell surface markers is one potential means of isolating stem cells. This method has been widely adopted to determine the presence of GSCs in heterogeneous tumor cells (18, 24, 25). One of the limitations of this assay is the lack of universal and definitive markers for GSCs. CD133 is the most common antigen and AC133 is a monoclonal antibody against this cell surface protein (24). This antibody appears to correlate to some extent with the presence of GSCs in multiple different cell populations (24, 37, 38). However, several studies have reported the presence of CD133-negative GSCs in GBM samples possibly due to patient-to-patient phenotypic and genetic differences (27). More recently, CD15 has gained attention as another potential marker for GSCs (38). At this point, any of the currently available cell surface antibodies are only surrogate markers for GSCs and need to be tested on a case-by-case basis. In the samples of this study, CD133, but not CD15, correlated with enrichment of GSCs (data not shown).

Telomestatin interacts with the G-quadruplex recognizing sequences in the genome which are present in the telomeric DNA (3), as well as the promoter regions of several proto-oncogenes (8–11). Previous studies have identified several potential target molecules including hTERT (13, 32), c-Myc (10), HIF1-α (11), RET (8), and platelet-derived growth factor receptor (PDGFR)-β (9). The effective doses found in most of these studies were relatively high, from around the several micromolar range. In this study, patient-derived GBM sphere cells underwent apoptosis at approximately 0.5 to 1 μmol/L, and no obvious change in expression of the above genes was detected. Instead, c-Myb, which also has the telomestatin-binding sequences in its promoter region (39), was the only gene showing a significant reduction in expression (Fig. 5). Both reduction of c-Myb expression and foci formation in telomeric DNA occurred in GBM spheres but not in non-GSCs following telomestatin treatment. Together, two distinct mechanisms of telomestatin action in GBM include, but may not be exclusive to, the direct disruption of telomere structure and inhibition of c-Myb.

In the CNS, c-Myb expression is restricted to the germinal zones of the adult brain and maintains progenitor cell proliferation and ependymal cell integrity (22). However, mice with the conditional brain-specific deletion of c-Myb exhibit only a mild defect in brain structure with hydrocephalus and ependymal cell abnormalities (22). To date, its function in gliomagenesis and maintenance of the immature state of GSCs has not been uncovered. We found that c-Myb expression is markedly elevated in clinical samples of GBM compared with normal brain tissues and undifferentiated GBM spheres compared with normal spheres or non-GSCs. Our functional data using knockdown and overexpression (Fig. 6) provide the first evidence that c-Myb plays a fundamental role in the growth of GSCs both in vitro and in vivo. Taken together, these data indicate that c-Myb is a potential new therapeutic target in the treatment of GBMs.

Telomestatin is a hydrophobic agent with a relatively large molecular weight (molecular weight = 583 kDa). The pharmacokinetics of this compound is not well defined and, more importantly, it is still unknown whether telomestatin penetrates the blood–brain barrier. In this study, we sought to use local (intratumoral) injection of telomestatin to deliver the drug directly to tumor cells. One of our aims is to identify safe and effective anti-GBM agents that can be applied to the removal cavity after tumor resection during surgery. Given that GBM is a tumor with an aggressive infiltrating nature, migrated tumor cells into the adjacent normal brain tissue are left behind during surgery. The presence of residual tumor cells makes relapse of the tumor inevitable within months or years of the surgery, despite adjuvant therapies. To achieve selective and effective eradication of the remaining tumor cells with minimal organ toxicity, we sought to identify anti-GSC agents that can be directly injected into the tumor cavity. Despite our positive data (Fig. 2), several open questions still remain. For instance, we cannot completely exclude the possibility of normal cell death following local injection of telomestatin. Particularly, its effect on long-term self-renewal of normal NSCs in the subventricular zone and hippocampus in the adult brain needs to be fully addressed.

In this study, we showed that telomestatin treatment preferentially targets GBM cells with stem cell–like characteristics. Telomestatin decreased sphere-forming potential in vitro, reduced CD133-positive cell population in vitro, and decreased tumor sizes in vivo. In addition, normal neural precursors treated with the same dose of telomestatin showed less cytotoxicity. A search for target genes of telomestatin by cDNA microarray identified the proto-oncogene c-Myb and further functional analysis using GBM spheres showed that c-Myb plays a critical role in the growth of GSCs both in vitro and in vivo. Finally, a significant elevation of c-Myb expression in surgical specimens of GBM supports the clinical relevance of our findings. We believe that these data serve as a rationale for the design of a novel anti-GBM therapy. Future studies will shed light on the molecular functional role of c-Myb in GBMs and the therapeutic window of telomestatin for GBMs.

All authors have seen and approved the manuscript. No potential conflicts of interest were disclosed.

The authors thank E.A. Chiocca (OSU), C.H. Kwon (OSU), J.N. Rich (Cleveland Clinic), P.J. Houghton (Nationwide Children's Hospital), P.S. Mischel (UCLA), and H.I. Kornblum (UCLA) for discussions and comments, help with materials, and constructive criticisms. They also thank Mr. Anthony Harrington for the expert editorial assistance.

This work was supported by the start-up fund from the OSU, the Department of Neurological Surgery, and the Khan Family Foundation.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.