Gliomas remain one of the deadliest forms of cancer. Improved therapeutics will require a better understanding of the molecular nature of these tumors. We, therefore, mimicked the most common genetic changes found in grade III-IV gliomas, disruption of the p53 and RB pathways and activation of telomere maintenance and independence from growth factors, through the ectopic expression of the SV40 T/t-Ag oncogene, an oncogenic form of H-ras (H-rasV12G), and the human telomerase catalytic subunit hTERT in normal human astrocytes. The resulting cells displayed many of the hallmarks of grade III-IV gliomas, including greatly expanded life span and growth in soft agar and, most importantly, were tumorigenic with pathology consistent with grade III-IV neuroectodermal tumors in mice. This model system will, for the first time, allow the biological significance of selected genetic alterations to be studied in human gliomas.
Brain tumors present one of the most severe challenges to oncology. Despite maximal therapy, the most common form of primary brain tumor, the glioblastoma multiforme is associated with a 5-year survival in <2% of glioblastoma multiforme patients (1). Therapeutic advances may be facilitated by an improved understanding of the molecular alterations involved in brain tumor formation and growth. Recent model systems have been used for studying the transformation of murine glia (2, 3, 4, 5), but genetic changes capable of transforming murine cells are insufficient to transform human cells (6), which suggests that human cells require different or additional genetic alterations to become transformed. We sought to create a genetically malleable model of transformation of the presumed cell of origin of the most prevalent primary brain tumors, the astrocyte, based on abnormalities detected frequently in human gliomas. The majority of human gliomas appear to have a set of pathways that are disrupted (the p53 and RB pathways) and others that are abnormally active (ras and telomerase; reviewed in Refs. 7, 8), which act together to permit glioma cells to acquire characteristics essential for tumor formation and growth (reviewed in Ref. 9). Alterations in the p53 and RB signal transduction pathways occur at some point in the majority of high-grade gliomas (7, 8, 10, 11) and permit cells to avoid apoptosis and growth-inhibitory signals. Many gliomas escape limitations of replication through the activation of the enzyme telomerase (12), which maintains the length of the telomeres. Gliomas, like most cancers, develop independence from external growth signals through the expression of growth factors or alteration of growth factor receptors or the downstream mediators. Ras is one such mediator of growth factor pathways and is activated in most high-grade gliomas (13), despite the absence of ras mutations, possibly through alterations in upstream growth factors and their receptors (14). Therefore, to create a genetically tractable system of astrocyte transformation, we used genetic alterations to target these pathways. Two recent studies (15, 16) have shown that human fibroblast, mammary, and kidney epithelial cells could be transformed through the introduction of the SV40 T/t-Ag,3 the human telomerase catalytic subunit hTERT, and an oncogenic form of H-ras (H-rasV12G). T-Ag has a variety of effects important in its transformation ability, but the two dominant ones are its ability to bind and inactivate two tumor suppressor genes, p53 and pRb-105 (reviewed in Ref. 17). The T-Ag construct also encodes t-Ag, which is vital to T-Ag function, possibly through the stabilization of the T-Ag-p53 complex or alteration of protein phosphatase 2A activity (17). Of note, T-Ag sequences have been purified from gliomas (17, 18). hTERT encodes the catalytic subunit of telomerase, which acts to maintain telomere length and permit indefinite replication in many cell types (19). H-rasV12G encodes an oncogenic form of the GTPase H-ras with decreased GTPase activity, activating mechanisms favoring cellular growth and invasion, angiogenesis, and immunosuppression (reviewed in Ref. 20). Given that T/t-Ag, hTERT, and H-rasV12G are tumorigenic in different cell types and that they affect pathways commonly altered in gliomas, we used T/t-Ag, hTERT, and H-rasV12G to transform astrocytes. We showed that human astrocytes that express T/t-Ag, H-rasV12G, and ectopic hTERT display key phenotypes of grade III-IV neuroectodermal tumors, such as expansion of life span and growth in soft agar, and are tumorigenic. Thus, we have created a system by which specific genetic alterations can be generated in human astrocytes or their precursor cells to study the biological consequences of particular molecular changes in malignant neuroectodermal tumor formation.
Materials and Methods
Generation of Cell Lines.
Normal human astrocytes (lot 16877) were purchased from Clonetics (Walkersville, MD). We confirmed the astrocytic nature of the cells by GFAP immunohistochemistry, with >95% cells staining positive. Cells were grown on plates coated with 13.3 μg/ml of poly-d-lysine (Sigma Chemical Co., St. Louis, MO) in astrocyte growth medium (Clonetics). Amphotrophic retrovirus for T/t-Ag, hTERT, and H-rasV12G was prepared as described previously (15). Retroviral constructs were introduced serially, and after infection, cells with successful viral integration were selected for positively by the presence of antibiotic resistance. Cells were selected in G418 (400 μg/ml; 7–10 days), hygromycin (100 μg/ml; 7–10 days), or puromycin (1 μg/ml; 5–7 days). Retroviruses carrying only the drug resistance gene were used as vector controls, except in the case of the hTERT, for which a hemagglutinin-labeled hTERT gene was used. This gene encodes a protein that retains in vitro telomerase activity but no in vivo activity (21). When cells reached confluence after the last viral infection, this was designated as population doubling 0. Cells were passaged either 1:4 (without T/t-Ag infection) or 1:16 (with T/t-Ag infection).
Cells were plated at a density of 5–10 × 103 cells/well in eight-well slides and grown to near confluence. Cells were fixed in formaldehyde and blocked in PBS with 0.1% saponin (Sigma Chemical Co.) and 10% normal goat serum (Zymed, South San Francisco, CA). Cells were treated with either an IgG control antibody or an anti-GFAP antibody mixture (both generous gifts of Carol Wikstrand) at 25 mg/ml. Cells were then treated with an antimouse IgG (Sigma Chemical Co.), horseradish peroxidase-strepavidin, and 3,3′-diaminobenzidine stain. Slides were counterstained with hematoxylin.
Cells were analyzed for the expression of T-Ag and ras by Western analysis. A 10-cm plate was lysed, and 60 μg of total cellular protein were used for each sample. Samples were subjected to SDS-PAGE analysis and transferred to a polyvinylidene difluoride membrane. The membrane was blocked in Tris-buffered saline with 0.05% Tween 20 and 5% albumin. Primary antibodies for anti-T-Ag (Santa Cruz Biotechnology, Santa Cruz, CA), anti-actin (Santa Cruz Biotechnology), or anti-ras (Calbiochem, La Jolla, CA) were used. Secondary antibodies were either goat antirabbit (Bio-Rad, Hercules, CA) or sheep antirabbit antibodies (Amersham, Piscataway, NJ).
RT-PCR and Telomerase Assays.
Growth Inhibition Assays.
Cells were plated into 12-well plates at a density of 2 × 104 cells/well and labeled for the last 6 h with 4 μCi of [3H]thymidine, fixed in 10% trichloroacetic acid, and lysed in 0.2 n NaOH. [3H]Thymidine incorporation into the DNA was measured with a scintillation counter. Each measurement was performed in triplicate.
Soft Agar Assays.
For these assays, 35-mm plates were prepared with a base layer of DMEM with 10% calf serum (Life Technologies, Inc., Rockville, MD) and 0.6% Bacto agar (Becton Dickinson and Co., Franklin Lakes, NJ). Cells were plated at a density of 5 × 104 cells/plate in a mix of DMEM with 10% calf serum and 0.4% Bacto agar. Plated cells were fed once a week with 0.5 ml of DMEM plus 10% calf serum and 0.4% Bacto agar. After 14 days, the plates were stained with 0.5 ml of 0.005% crystal violet. On each plate, colonies with >30 cells were counted. Each measurement was performed in triplicate.
Tumor Formation Assays.
scid-beige mice (Taconic, Germantown, NY) were s.c. injected in the flank with 10 × 106 cells/mouse in 100 μl of Matrigel (Becton Dickinson and Co.; Ref. 23). Mice were regularly checked for tumor formation. Tumor volume was calculated with the formula 0.5 × (length) × (width)2. Tumors were removed and examined by immunohistopathology and electron microscopy. Intracranial tumor formation was tested with nude athymic BALB/c mice radiated with 500 rads injected with 1 × 106 cells in 50 μl of Matrigel. Mice were sacrificed when they developed any behavioral abnormalities, and brains were examined by histopathology.
Creation of Genetically Modified Human Glial Cell Lines.
Human astrocytes stably expressing T/t-Ag, hTERT, and H-rasV12G were created with vector controls through the use of independently selectable markers (neomycin, hygromycin, and puromycin, respectively) expressed by retroviral constructs. We confirmed overexpression of T-Ag and ras by Western analysis (Fig. 1,A), hTERT expression by RT-PCR specific for ectopic hTERT RNA (Fig. 1,B), and restoration of in vitro telomerase activity by TRAP assay (Fig. 1 C). All cell lines had in vitro telomerase activity, because they expressed either hTERT or HA-hTERT. HA-hTERT was used as a control because it exhibits in vitro telomerase activity but lacks in vivo activity (21). GFAP was detectable in all of the infected cells but was of lesser intensity than in the original cells (data not shown). Cells infected with T/t-Ag developed a more compact cellular morphology, and cells expressing hTERT alone retained a large flattened morphology (data not shown).
Expression of T/t-Ag and hTERT Extends Life Span.
Expression of T/t-Ag and hTERT each increased DNA synthesis as measured by thymidine incorporation, whereas expression of H-rasV12G caused no such increase (data not shown). Passage of astrocyte cell lines revealed that vector control cells senesced, and cells expressing hTERT alone had a moderately extended life span but were not immortal (Fig. 2,A). Expression of H-rasV12G in the absence of T/t-Ag expression, with or without the expression of hTERT, caused rapid cellular senescence, as expected (24). Both hTERT + T/t-Ag and T/t-Ag cells grew well past senescence to >200 population doublings (Fig. 2,A). Long-term passage of cells has been shown to be associated with induction of endogenous telomerase activity (25). Therefore, we examined early passage cells (<16 population doublings) and found an absence of endogenous hTERT expression (Fig. 2 B). However, analysis of late passage cells expressing T/t-Ag (>150 population doublings) revealed endogenous hTERT expression (data not shown) as a probable explanation for continued division. We therefore used early-passage cells in all experiments to minimize unintended genetic alterations.
T/t-Ag, hTERT, and H-rasV12G Transform Human Astrocytes.
Anchorage-independent growth of our neuroectodermal cell lines at early passage was measured through colony formation in soft agar. Colony formation in soft agar represents one of the best in vitro measurements of cellular transformation (reviewed in Ref. 26) because this assay tests for characteristics essential to cancer formation, including clonal growth as well as growth independent of both cellular and extracellular matrix interactions. Only cells expressing T/t-Ag, hTERT, and H-rasV12G formed a significant number of colonies in soft agar (Fig. 3 A).
As an in vivo measure of cellular transformation, we assayed the ability of our cell lines to form progressively growing tumors when implanted s.c. in immunocompromised scid-beige mice (Fig. 3 B). In results parallel to the soft agar colony formation assays, only astrocytes expressing T/t-Ag, hTERT, and H-rasV12G formed tumors in mice. s.c. tumor formation was tested to ensure the maximal access to tumor tissue and allow measurement of tumor growth; however, we found that only the cells expressing T/t-Ag, hTERT, and H-rasV12G also formed intracranial tumors in radiated nude athymic mice (data not shown). Tumors grown s.c. became palpable at 5 weeks after cellular implantation, grew rapidly in size, and were serially passaged. Cells derived from the xenografts followed similar growth kinetics when reimplanted s.c. in scid-beige mice (data not shown), which indicated that there was no selection of a subpopulation of cells with specific characteristics with a growth advantage.
Microscopic review of the s.c. tumors revealed malignant features including high nuclear:cytoplasmic ratios, with the large pleomorphic nuclei demonstrating speckled chromatin, prominent nucleoli, and frequent mitotic figures. The cytoplasm of these cells was frequently drawn into elongated cytoplasmic profiles in areas of looser growth, whereas in more compact areas of growth, the tumor exhibited an epithelioid growth pattern. The vascularity was prominent, although no endothelial vascular proliferation was evident. Necrosis, although not a common feature of the tumors, could be found and was associated with accumulations of tumor cells around the borders of necrosis (Fig. 4,A). Immunohistochemically, the tumor cells were strongly immunoreactive for keratin and S100 (Fig. 4,B), consistent with a neuroectodermal origin. Staining for GFAP, desmin, and smooth muscle was nonreactive. Although the cells used for injections expressed GFAP, the absence of GFAP staining of our tumors is not surprising; this is common for grade III-IV gliomas grown as xenografts and particularly with cells that have been serially passaged in cell culture (27). The frequent mitotic figures correlated with a high Ki-67 labeling index of ∼50% (Fig. 4 C). Ultrastructural examination also revealed features of a high-grade tumor with significant nuclear pleomorphism and a simplified organellar pattern without specialized differentiation. Specifically, intermediate filaments, neurosecretory granules, basal lamina, and cellular junctions were not found, and there were no subplasmalemmal densities. There were no features consistent with a fibroblast cellular origin. The tumors formed thus represent a poorly differentiated, grade III-IV malignant neuroectodermal tumor.
Improved treatment of patients with primary brain tumors will require a significant expansion of our knowledge of mechanisms underlying the formation of brain tumors. We have now developed a novel system for elucidating the molecular etiology of primary human gliomas. From an initial culture of human astrocytes we have made defined genetic alterations, the expression of the oncogenes T/t-Ag and H-rasV12G with ectopic hTERT expression, to create cells that extend their life span long beyond senescence and acquire a tumorigenic phenotype, as measured by anchorage-independent growth in soft agar and tumor formation in mice. The genetic alterations that we used to convert normal astrocytes into malignant neuroectodermal tumors, although not directly representative of changes found in all human gliomas, disrupt or activate pathways that are similarly affected in human gliomas—p53, RB, telomere maintenance, and ras (7, 8, 10, 11, 12, 13, 14).
This model system is the first to allow for the genetic alteration of human astrocytes to create a model of glioma formation. However, it is not without its limitations:
(a) The expression of an oncogenic ras with the inactivation of the p53 pathway by T/t-Ag dramatically increases genomic instability (28). Any genetically tractable system using these elements must be tested at an early stage to ensure that the biological relevance of a genetic alteration is being measured and not an eventual, secondary effect. One example of a potentially confounding alteration that we found in late-passage cells expressing T/t-Ag was the induction of endogenous hTERT expression, complicating the interpretation of the role of T/t-Ag in the immortalization of our cells. To overcome this limitation, we assayed only early-passage cells.
(b) T-Ag is not involved in the majority of brain tumors; however, the use of T-Ag in our model does have unique biological relevance because of the potential role of T-Ag in brain tumor formation (17, 18), possibly because of exposure to contaminated polio vaccine (17).
(c) Mutations of ras are quite infrequent in human gliomas, but ras activity is frequently increased in grade III-IV gliomas (13). Future studies will involve the substitution of other mitogenic genetic changes in the place of ras.
(d) Our tumor cells lacked most specialized differentiation features of astrocytes, although they were S100 positive. Because our untransformed cells contained >95% GFAP-positive astrocytes and our preimplantation infected cells were still GFAP positive, we believe that we have transformed astrocytes. It is possible that we transformed neuroectodermal precursor cells that were in the small fraction of GFAP-negative cells in the original culture. Nevertheless, whether we transformed astrocytes that have shut off some differentiation features or neuroectodermal “stem” cells, the resultant tumor is a very high grade III-IV malignant neuroectodermal tumor. As stated in Russell and Rubinstein’s, “that the glioblastoma is primarily derived from astrocytes in most cases can be assumed, although its morphology may be so bizarre as to obscure such recognition” (29).
Rapid improvements in the measurement of gene expression are yielding a wealth of data in the study of cancer, but these advances lack the ability to determine the biological relevance of specific molecular alterations in human cells. No genetically tractable model of human glioma formation currently exists. Thus, the development of a genetically tractable model of human astrocyte transformation permits linkage between gene expression and changes in cellular behavior. Future studies will elucidate the contribution of other molecular alterations specifically found in human gliomas. Our model system will allow for the expression of additional genes in lieu of or in addition to T/t-Ag, H-rasV12G, and hTERT and the determination of the biological consequences. Additionally, this system provides a potential in vivo therapeutic model to study therapies directed at essential pathways in glioma formation and growth.
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.
Supported by NIH Grant K08 NS02055, an American Cancer Society Duke University Institutional Research Grant, and the Pediatric Brain Tumor Foundation (all to J. N. R.) and the Sidney Kimmel Foundation, the V-Foundation, and NIH Grants ROI CA82481 (to C. M. C.) and ROI CA83770 (to X-F. W.).
The abbreviations used are: T-Ag, large T antigen; t-Ag, small t antigen; GFAP, glial fibrillary acidic protein; RT-PCR, reverse transcription-PCR; TRAP, telomeric repeat amplification protocol; scid, severe combined immunodeficient.
We thank Qing Shi, Yong Yu, Steven Keir, and Allyson Smith for technical assistance. Drs. Thomas Cummings, Carol Wikstrand, Nesrin Hamad, and Yueyi Liu provided assistance in these studies. Janet Parsons provided editorial assistance.