A human brain tumor-derived PDGFR-alpha deletion mutant is transforming.

Aberrant receptor tyrosine kinase signaling plays an important role in the molecular pathogenesis of brain tumors. We have been studying a previously identified human glioblastoma-derived PDGFR-alpha mutant that has an in-frame deletion in the extracellular domain, causing loss of exons 8 and 9 (PDGFR-alpha(delta8,9)). In the primary tumor, this deletion mutant receptor was shown to be amplified and overexpressed. The purpose of this study was to determine the expression, activity, localization, and transformation properties of this deletion mutant. In the absence of serum, or PDGF-AA, PDGFR-alpha(delta8,9) was phosphorylated on tyrosine residues, indicating ligand-independent autoactivation. Localization by staining and cell surface biotinylation studies revealed expression of the deletion mutant predominantly in the cytoplasm, with very little present on the cell surface. To determine if PDGFR-alpha(delta8,9) was oncogenic, we transfected wild-type and mutant receptors into Rat1 cells and performed analyses of cell growth, in vitro transformation, and subcutaneous growth in the nude mouse. PDGFR-alpha(delta8,9)-expressing cells displayed enhanced cell growth and survival in low serum, and formed foci in monolayer cultures. PDGFR-alpha(delta8,9)-expressing Rat1 cells were also tumorigenic when injected subcutaneously into nude mice. Expression of PDGFR-alpha(delta8,9) was also associated with increased c-Jun phosphorylation in the absence of PDGF ligand, demonstrating also that the mutant receptor is associated with altered intracellular signaling. These data demonstrate that PDGFR-alpha(delta8,9) is transforming, and it is the first demonstration of a naturally occurring tumor-derived mutant PDGFR-alpha with oncogenic properties.

Glioma migration: clues from the biology of neural progenitor cells and embryonic CNS cell migration.

Neural stem cells have recently come to the forefront in neurobiology because of the possibilities for CNS repair by transplantation. Further understanding of the biology of these cells is critical for making their use in CNS repair possible. It is likely that these discoveries will also have spin-offs for neuro-oncology as primary brain tumors may arise from a CNS progenitor cell. An understanding of the normal migratory ability of these cells is also likely to have a very important impact on the knowledge of brain tumor invasion.

Cip/Kip cell-cycle inhibitors: a neuro-oncological perspective.

The cell cycle is a precisely controlled cellular program that ensures normal cellular proliferation and development. The cyclin-dependant kinases (CDK) are molecules central to the continued progression through the cell-cycle checkpoints and as such are regulated by various mechanisms including cyclin levels, phosphorylation/dephosphorylation and cyclin-dependant kinase inhibitors (CKI). The CKIs are grouped into two families based on their structure and function, four lnk4 CKIs and three Cip/Kip CKIs. Abnormalities in these proteins can give rise to developmental defects and cancer. In this review, we will discuss the biochemistry and cell biology of the each of the Cip/Kip CKIs, their role in development as evidenced by targeted mutations in mice, and their role as possible tumor suppressor genes.

The INK4A/ARF locus: role in cell cycle control and apoptosis and implications for glioma growth.

The unique INK4A/ARF locus at chromosome 9p21 encodes two distinct proteins that intimately link the pRB and p53 tumour suppressor pathways. p16INK4A has been identified as an inhibitor of the cell cycle, capable of inducing arrest in G1 phase. p14/p19ARF on the other hand can induce both G1 and G2 arrest due to its stabilizing effects on the p53 transcription factor. In addition to their roles in growth arrest, both proteins are involved in cellular senescence and apoptosis. The frequent mutation or deletion of INK4A/ARF in human tumours as well as the occurence of tumours in the murine knockout models have identified both p16 and ARF as bona fide tumour suppressors.

Cytoskeletal-induced alterations in the adhesion of HT-1080 fibrosarcoma cells to extracellular matrix.

We have investigated the adhesion of the human fibrosarcoma cell line, HT-1080, transfected with glial fibrillary acidic protein (GFAP) to a variety of extracellular matrix macromolecules (ECM) including collagen type IV, laminin, and fibronectin. The GFAP-transfectants demonstrated altered adhesiveness to extracellular matrix substrates when compared to controls. GFAP-positive, heavy metal-induced fibrosarcoma cells were more adherent to plastic and collagen type IV than were the parental or uninduced cells. In contrast, GFAP-positive fibrosarcoma cells were less adherent to laminin- or fibronectin-coated dishes than controls. Time course adhesion studies over 9 days showed that the heavy metal-induced fibrosarcoma cells progressively became more adherent to collagen type IV and less adherent to laminin- or fibronectin-coated dishes than did uninduced cells. However, with the removal of heavy metal from the medium, the HT-1080 fibrosarcoma cells were restored to their original adhesive potential. By phase microscopy, uninduced and induced HT-1080 cells demonstrated different morphological features and remained viable in an anchorage-dependent fashion on collagen type IV as a substrate. By way of contrast, GFAP-induced HT-1080 cells were not particularly viable in monolayer culture and readily detached from laminin as a substrate. The expression of beta1 integrin in GFAP-positive fibrosarcoma cells was decreased following heavy metal induction by Western blot analyses. In contrast, the expression of alpha2 integrin was increased whereas alpha5 integrin was unchanged in HT-1080 cells following the induction of GFAP. Gelatin zymography showed that 72 kDa collagenase was less expressed in GFAP-induced clones than in controls. Our data suggest that the forced expression of the intermediate filament, GFAP, in HT-1080 cells may modulate cell adhesion to different ECM substrates through alterations in expression of integrins.

Expression of p57(KIP2) potently blocks the growth of human astrocytomas and induces cell senescence.

Astrocytic tumors frequently exhibit defects in the expression or activity of proteins that control cell-cycle progression. Inhibition of kinase activity associated with cyclin/cyclin-dependent kinase co-complexes by cyclin-dependent kinase inhibitors is an important mechanism by which the effects of growth signals are down-regulated. We undertook the present study to determine the role of p57(KIP2) (p57) in human astrocytomas. We demonstrate here that whereas p57 is expressed in fetal brain tissue, specimens of astrocytomas of varying grade and permanent astrocytoma cell lines do not express p57, and do not contain mutations of the p57 gene by multiplex-heteroduplex analysis. However, the inducible expression of p57 in three well-characterized human astrocytoma cell lines (U343 MG-A, U87 MG, and U373 MG) using the tetracycline repressor system leads to a potent proliferative block in G(1) as determined by growth curve and flow cytometric analyses. After the induction of p57, retinoblastoma protein, p107, and E2F-1 levels diminish, and retinoblastoma protein is shifted to a hypophosphorylated form. Morphologically, p57-induced astrocytoma cells became large and flat with an expanded cytoplasm. The inducible expression of p57 leads to the accumulation of senescence-associated beta-galactosidase marker within all astrocytoma cell lines such that approximately 75% of cells were positive at 1 week after induction. Induction of p57 in U373 astrocytoma cells generated a small population of cells ( approximately 15%) that were nonviable, contained discrete nuclear fragments on Hoechst 33258 staining, and demonstrated ultrastructural features characteristic of apoptosis. Examination of bax and poly-(ADP ribose) polymerase levels showed no change in bax, but decreased expression of poly-(ADP ribose) polymerase after p57 induction in all astrocytoma cell lines. These data demonstrate that the proliferative block imposed by p57 on human astrocytoma cells results in changes in the expression of a number of cell cycle regulatory factors, cell morphology, and a strong stimulus to cell senescence.

Imaging correlates of successful endoscopic third ventriculostomy.

OBJECT: The goal of this study was to determine and compare imaging correlates in pediatric patients who underwent successful or failed endoscopic third ventriculostomies (ETVs). To this end, the authors measured ventricular size changes and the presence of cerebrospinal fluid (CSF) flow void in both groups of children following ETV. METHODS: Images obtained in children with hydrocephalus immediately before and at least 30 days after having undergone ETV were reviewed by four independent observers (two blinded and two nonblinded). Each observer independently measured the frontal and occipital horn ratio ([FOR], a reliable and valid measure of ventricular size) and provided a subjective assessment of the presence of a flow void at the ETV site, the degree of periventricular edema, and the amount of CSF over the cerebral hemispheres. There were 29 children whose mean age was 6.6 years at the time of ETV and who had a mean postoperative follow-up period lasting 1.6 years. Postoperatively, the mean reduction in ventricular size (as measured using the FOR) was 7% (95% confidence interval [CI] 3-11%) in cases that were deemed failures (eight patients) and 16% (95% CI 12-20%) in clinically successful cases (21 patients). This reduction was significantly greater in cases of clinical success compared with those that were deemed failures (p = 0.03, t-test). There were no substantial differences between blinded and nonblinded assessments. Flow void was present in 94% of successes and absent in 75% of failures (p = 0.01, Fisher's exact test). The other subjective assessments were not significantly different between the groups of successes and failures. CONCLUSIONS: Ventricular size appears to be somewhat reduced in both groups of patients who underwent clinically successful and failed ETV; however, the reduction is significantly greater among clinically successful cases. The presence of a flow void also appears to correlate with clinical success and its absence with clinical failure.

Measurement of ventricular size: reliability of the frontal and occipital horn ratio compared to subjective assessment.

INTRODUCTION: The frontal and occipital horn ration (FOR) has recently been described as a simple, linear measurement of ventricular size that correlates very well with ventricular volume. This study further characterizes the measurement properties of the FOR by investigating its interobserver reliability and comparing it to a subjective assessment of ventricular size. METHODS: Axial images (CT and MR) of children with hydrocephalus taken before and after third ventriculostomy were reviewed by 4 independent observers. Two observers were blinded to patient identity and clinical status and 2 observers were nonblinded. Each observer independently recorded linear measurements from which the FOR was calculated for each image. Each reviewer also made a separate subjective assessment of the degree of hydrocephalus on a 9-point adjectival scale. Reliability was calculated using a repeated-measures analysis of variance (ANOVA) and an intraclass correlation coefficient (ICC) with random image and observer effects. RESULTS: There were 120 separate observations (4 observers, 30 images). The FOR ranged from 0.33 to 0.75 (mean 0.55, standard deviation 0.11). The reliability coefficient was 0.93 (95% confidence interval, CI 0.80-0.97) between the 2 blinded observers and 0.98 (95% CI, 0.95-0.99) between the 2 nonblinded observer. The overall interobserver reliability for all 4 observers was 0.95 (95% CI 0.92-0.98). The mean FOR for each observer was very similar, regardless of the observer's blinding status. However, the reliability of the observers' subjective assessment of the hydrocephalus was much lower (ICC = 0.77, 95% CI 0. 60-0.88). CONCLUSIONS: The FOR demonstrates excellent interobserver reliability (>0.9) and was superior to subjective assessments of hydrocephalus. In this study, excellent reliability was maintained regardless of the blinding status of the observers. This further demonstrates the properties of the FOR as a simple and reproducible measure of ventricular size. It is suitable for use in clinical studies, possibly even in situations in which observer blinding is not possible.

Astrocytoma adhesion to extracellular matrix: functional significance of integrin and focal adhesion kinase expression.

Evidence is accumulating implicating a role for integrins in the pathogenesis of cancer, a disease in which alterations in cellular growth, differentiation, and adhesive characteristics are defining features. In the present report we studied a panel of 8 human astrocytoma cell lines for their expression of integrin subunits by RT-PCR, and of integrin heterodimers by immunoprecipitation analyses. The functionality of integrin heterodimers was assessed using cell attachment assays to plastic or single matrix substrates. Downstream effects of integrin activation were studied by western blot analyses of FAK expression in human astrocytoma cell lines growing on plastic and on a fibronectin matrix, and in 13 primary human brain tumor specimens of varying histopathological grade. Furthermore, we studied tyrosine phosphorylation of FAK in astrocytoma cells growing on plastic versus fibronectin. Finally, we analyzed the effects of intermediate filament gene transfer on FAK phosphorylation in SF-126 astrocytoma cells. Our data show that astrocytoma cell lines express various integrin subunits by RT-PCR, and heterodimers by immunoprecipitation analyses. The beta1 and alphav integrin subunits were expressed by all astrocytoma cell lines. The alpha3 subunit was expressed by all cell lines except SF-188. By immunoprecipitation, the fibronectin receptor (alpha5beta1 integrin heterodimer) and the vitronectin receptor (alphavbeta3) were identified in several cell lines. Astrocytoma cell attachment studies to human matrix proteins suggested that these integrin heterodimers were functional. Using confocal laser microscopy, we showed that FAK was colocalized to actin stress fibers at sites of focal adhesion complexes. By western blot, FAK was variably but quite ubiquitously expressed in human astrocytoma cell lines, and in several primary human astrocytoma specimens. When U373 and U87 MG astrocytoma cells bind to a fibronectin matrix, FAK is phosphorylated. GFAP-transfected SF-126 human astrocytoma cells were shown to overexpress the phosphorylated form of FAK only when these cells were placed on a fibronectin matrix. This result is of interest because it suggests that manipulations of the astrocytoma cytoskeleton per se can bring about potential signaling changes that channel through integrins and focal adhesion sites leading to activation of key kinases such as FAK.

The E2F-family proteins induce distinct cell cycle regulatory factors in p16-arrested, U343 astrocytoma cells.

We previously demonstrated that P16Ink4a (p16) expression in p16-deficient U343 astrocytoma cells causes a G1 cell cycle arrest, profound changes in cytoskeletal proteins and alterations in expression and activity of the pRB and E2F family proteins. We examine here the effects of expressing wild type or mutant versions of the downstream targets of p16 in U343 astrocytomas. We first attempted to block proliferation of U343 cells using the dominant mutant of pRB, deltap34. Expression of this mutant in the human osteosarcoma, SAOS-2, potently blocked proliferation but did not affect the cell cycle of U343 cells. We next showed that expression of E2F-1, E2F-2, E2F-3 and E2F-4 are each able to overcome this p16-dependent cell cycle arrest but exhibit distinct biological activities. Adenoviral-mediated expression of E2F-1, E2F-2, E2F-3, or E2F-4 overcame the p16-dependent cell cycle block and induced alterations in cell morphology. E2F-5, only in conjunction with DP1, promoted cell cycle progression. For both E2F-1 and E2F-2, but not E2F-3 or E2F-5/DP1, cell cycle re-entry was associated with almost quantitative cell death. Only small numbers of dying cells were observed in E2F-4-expressing cultures. Expression of the different E2F's altered the expression of distinct sets of cell cycle regulatory proteins. E2F-1 induced endogenous E2F-4 expression and also caused an increase in pRB, p107 and cyclin E levels. Expression of E2F-4 caused a weak increase in E2F-1 levels but also strongly induced pRB, p107, p130 and cyclin E. However, E2F-1 and E2F-4 clearly regulate expression of distinct genes, demonstrated when E2F-4 caused a threefold increase in the levels of cdk2 whereas E2F-1 failed to increase in this cyclin dependent kinase. Similarly, expression of E2F-1 or E2F-2 were shown to have distinct effects on the expression of cdk2, cyclin E and pRB despite both of these closely related E2F-family members potently inducing cell death. Thus, E2F-1, E2F-2, E2F-3 and E2F-4 are able to overcome the p16-dependent proliferative block in U343 astrocytoma cells. While overcoming this cell cycle block, each of the E2F's uniquely affect the expression of a number of cell cycle regulatory proteins and have distinct abilities to promote cell death.

Characterization of glial filament-cytoskeletal interactions in human astrocytomas: an immuno-ultrastructural analysis.

The role that glial filaments play in cells and tumors of glial origin is not well understood. We therefore undertook the present study to determine the relationships between glial and vimentin intermediate filaments (IFs), actin microfilaments, and CD44, a cell surface glycoprotein important in cell migration and invasion, in human astrocytoma cells. Three astrocytoma cell lines, U343 MG-A (U343), U251 MG (U251), and antisense GFAP-transfected U251 (asU251) were studied using immunofluorescence confocal and immunoelectron microscopy. Furthermore, we studied the phenotypic behaviour of these astrocytoma cell lines by analyzing their migration through Matrigel in vitro. U343 astrocytoma cells had the highest expression levels of glial fibrillary acidic protein (GFAP), whereas asU251 had virtually no expression of GFAP. Parental U251 cells had intermediate expression levels of GFAP. The elimination of GFAP expression in as U251 cells was accompanied by a marked increase in vimentin, actin microfilaments and CD44 levels. Gold labeling density counts of cytoskeletal and cell surface elements demonstrated that the differences between GFAP, actin, CD44 and vimentin levels in the different astrocytoma cell lines were statistically significant (p < 0.05). Results from the in vitro invasion assay revealed that U343 cells demonstrated the least invasive potential, whereas asU251 astrocytoma cells demonstrated the most. Our results show that elimination of GFAP expression by antisense leads to marked alterations in cell morphology and phenotypic behaviour. These data imply that GFAP may be linked spatially and functionally to cytoskeletal elements which may be altered when this IF is deleted in astrocytomas.

Retinoic acid and the cyclin dependent kinase inhibitors synergistically alter proliferation and morphology of U343 astrocytoma cells.

We have characterized the expression and activity of the cell cycle regulatory machinery and the organization of the cytoskeleton of the p16(Ink4a)-deficient astrocytoma cell line, U343 MG-a (U343), following retinoic acid (RA) treatment. RA causes cell cycle arrest at low cell density and significant morphological changes in U343 cells, reflected by reorganization of the intermediate filament, GFAP, and actin. RA-induced cell cycle arrest is also associated with induction of p27Kip1 expression, inhibition of cdk2-associated kinase activity and alteration of the phosphorylation state of the pRB-family proteins. We next determined the effect of inducing expression of the cyclin dependent kinase inhibitors (CKI's), p16(Ink4a), p21Cip1/Waf1 or p27Kip1 on the proliferation and morphology of these malignant astrocytoma cells in the absence and presence of RA. Induction of p16, p21 or p27, using the tetracycline repressor system, potently inhibits proliferation of U343 cells. However, rather than resembling RA-treated cells, CKI-induced U343 cells become flat with abundant cytoplasm and perinuclear vacuolization. CKI-induced morphological alterations are accompanied by a significant reorganization of glial filaments within the cytoplasm. Interestingly, when U343 cells are growth arrested by p16, p21 or p27 induction and treated simultaneously with RA, a dramatic morphological change occurs, cells acquiring multiple long, tapering processes reminiscent of primary astrocytes. This rearrangement is accompanied by reorganization of GFAP, vimentin and actin. Vimentin specifically relocalizes to the tips of the long processes which form. The arrangement of intermediate filaments in these cells is, in fact, indistinguishable from their arrangement in primary human astrocytes. These data demonstrate that when a strong proliferative block, produced by CKI expression, occurs in conjunction with the morphogenic signals generated by RA, these p16-deficient malignant astrocytoma cells are induced to phenotypically resemble normal astrocytes.

Role of glial filaments in cells and tumors of glial origin: a review.

In the adult human brain, normal astrocytes constitute nearly 40% of the total central nervous system (CNS) cell population and may assume a star-shaped configuration resembling epithelial cells insofar as the astrocytes remain intimately associated, through their cytoplasmic extensions, with the basement membrane of the capillary endothelial cells and the basal lamina of the glial limitans externa. Although their exact function remains unknown, in the past, astrocytes were thought to subserve an important supportive role for neurons, providing a favorable ionic environment, modulating extracellular levels of neurotransmitters, and serving as spacers that organize neurons. In immunohistochemical preparations, normal, reactive, and neoplastic astrocytes may be positively identified and distinguished from other CNS cell types by the expression of the astrocyte-specific intermediate filament glial fibrillary acidic protein (GFAP). Glial fibrillary acidic protein is a 50-kD intracytoplasmic filamentous protein that constitutes a portion of, and is specific for, the cytoskeleton of the astrocyte. This protein has proved to be the most specific marker for cells of astrocytic origin under normal and pathological conditions. Interestingly, with increasing astrocytic malignancy, there is progressive loss of GFAP production. As the human gene for GFAP has now been cloned and sequenced, this review begins with a summary of the molecular biology of GFAP including the proven utility of the GFAP promoter in targeting genes of interest to the CNS in transgenic animals. Based on the data provided the authors argue cogently for an expanded role of GFAP in complex cellular events such as cytoskeletal reorganization, maintenance of myelination, cell adhesion, and signaling pathways. As such, GFAP may not represent a mere mechanical integrator of cellular space, as has been previously thought. Rather, GFAP may provide docking sites for important kinases that recognize key cellular substrates that enable GFAP to form a dynamic continuum with microfilaments, integrin receptors, and the extracellular matrix.

Current concepts in neuro-oncology: the cell cycle--a review.

Uncontrolled cellular proliferation is the hallmark of human malignant brain tumors. Their growth proceeds inexorably, in part because their cellular constituents have an altered genetic code that enables them to evade the checks and balances of the normal cell cycle. Recently, a number of major advances in molecular biology have led to the identification of several critical genetic and enzymatic pathways that are disturbed in cancer cells resulting in uncontrolled cell cycling. We now know that the progression of a cell through the cell cycle is controlled in part by a series of protein kinases, the activity of which is regulated by a group of proteins called cyclins. Cyclins act in concert with the cyclin-dependent kinases (CDKs) to phosphorylate key substrates that facilitate the passage of the cell through each phase of the cell cycle. A critical target of cyclin-CDK enzymes is the retinoblastoma tumor suppressor protein, and phosphorylation of this protein inhibits its ability to restrain activity of a family of transcription factors (E2F family), which induce expression of genes important for cell proliferation. In addition to the cyclins and CDKS, there is an emerging family of CDK inhibitors, which modulate the activity of cyclins and CDKs. CDK inhibitors inhibit cyclin-CDK complexes and transduce internal or external growth-suppressive signals, which act on the cell cycle machinery. Accordingly, all CDK inhibitors are candidate tumor suppressor genes. It is becoming clear that a common feature of cancer cells is the abrogation of cell cycle checkpoints, either by aberrant expression of positive regulators (for example, cyclins and CDKs) or the loss of negative regulators, including p21Cip1 through loss of function of its transcriptional activator p53, or deletion or mutation of p16ink4A (multiple tumor suppressor 1/CDKN2) and the retinoblastoma tumor suppressor protein. In this review, we describe in detail our current knowledge of the normal cell cycle and how it is disturbed in cancer cells. Because there have now been a number of recent studies showing alterations in cell cycle gene expression in human brain tumors, we will review the derangements in both the positive and negative cell cycle regulators that have been reported for these neoplasms. A thorough understanding of the molecular events of the cell cycle may lead to new opportunities by which astrocytoma cell proliferation can be controlled either pharmacologically or by gene transfer techniques.

Cyclin and cyclin-dependent kinase expression in human astrocytoma cell lines.

The expression of cyclins (A, B1, D1, D3, E), cyclin-dependent kinases (CDK2(3), CDK4), and the cyclin-dependent kinase inhibitors (CDKIs) p16(INK4A) and p21(CIP1) was studied in 9 malignant human astrocytoma cell lines using northern blot analysis, immunocytochemistry, and immunoblotting to see if their altered expression contributed to astrocytoma proliferation. Steady state cell cycle mRNA expression was analyzed in unsynchronized tumor cells, and cell cycle phase-specific gene expression was analyzed in 3 synchronized cell lines. Analysis of steady state expression revealed increased levels of several different cyclin transcripts and CDKs in a number of astrocytoma cell lines compared with normal human brain tissue or cultured fibroblasts. We confirm previous reports identifying loss of p16(INK4A) in astrocytomas, as a p16(INK4A) transcript was identified in only 2 cell lines and protein in 1 cell line. However, we now show that p21(CIP1) expression was also diminished relative to normal fibroblasts in all astrocytoma cell lines studied regardless of p53 mutation status. Analysis of synchronized astrocytoma cells revealed altered timing of mRNA expression of several cyclins. Immunocytochemistry revealed a generalized increase in immunoreactivity of astrocytoma cells for most cyclins and CDKs compared with human fibroblasts. Immunoblotting also revealed increased expression of cyclin proteins in a number of astrocytoma cell lines. These data suggest that increased expression of cyclins and CDKs, and decreased expression of CDKIs by human astrocytoma cell lines may contribute to their increased proliferative state. In addition, our data show that alterations in cell cycle genes in astrocytomas are not confined to the cyclin D1-CDK4-pRb axis.

Expression of stromelysin 1 in human astrocytoma cell lines.

In a wide variety of tumor types, the expression of stromelysin 1 which is one of the matrix metalloproteinases (MMPs) has been shown to correlate with tumor invasion. However, little is known about the distribution of stromelysin in human brain tumors. We have previously shown that a correlation exists between the type IV collagenases, tissue inhibitor of metalloproteinase (TIMP)-1 and TIMP-2 transcripts and in vitro invasiveness among 7 human astrocytoma cell lines. In the present study, we analyzed the expression of stromelysin 1 among the same panel of human astrocytoma cell lines and human fibroblasts by northern blot analyses and in situ hybridization. Northern blot analysis demonstrated that SF-126 and U87 MG expressed high level stromelysin 1 transcripts. Following heat shock stimulation, the stromelysin 1 transcript was up-regulated in U87 MG astrocytoma cells. In situ hybridization analysis showed specific intracytoplasmic localization of mRNA for stromelysin in these astrocytoma cell lines. By casein zymography, we have determined that both SF-126 and U87 MG secreted stromelysin 1 protein. We conclude that stromelysin 1 is expressed by certain human astrocytoma cell lines, and this study confirms the importance of continuing to characterize the proteolytic enzyme profile of these tumors to fully understand the molecular mechanisms involved in astrocytoma invasiveness.

Inducible expression of glial fibrillary acidic protein in HT-1080 human fibrosarcoma cells.

Glial fibrillary acidic protein (GFAP) is an intermediate filament protein expressed almost exclusively by glial cells of the central nervous system. We have previously transfected GFAP-negative human astrocytoma cells with the gene for GFAP and have demonstrated that GFAP transfection decreases astrocytoma proliferation and alters astrocytoma morphology. To determine if the same cellular responses could be elicited upon GFAP transfection of nonglial tumor cells, in the present study we have transfected a GFAP-negative human malignant fibrosarcoma cell line (HT-1080) with a cDNA containing the entire coding sequence of the human GFAP gene under the control of an inducible metallothionein promoter. Stably transfected HT-1080 clones were identified that are GFAP-positive by PCR and immunocytochemistry. GFAP-positive HT-1080 fibrosarcoma cells also demonstrate a decrease in tumor cell proliferation, altered morphological features characterized by cell elongation and cytoplasmic process formation, and reduction of invasive potential when compared to controls. These findings suggest that the inducible expression of the cytoskeletal protein GFAP can also be associated with dramatic cellular effects in nonglial non-central nervous system tumor cells.

Supratentorial primitive neuroectodermal tumors in children.

A retrospective review of 36 children diagnosed with a supratentorial primitive neuroectodermal tumor (PNET) at the Hospital for Sick Children was performed for the period 1970-1995. All children but one received their initial treatment at our institution. There were 18 males and 18 females and the median age at diagnosis was 35 months. Twenty-two PNETs were lobar, 3 were deep in the hemisphere, and 10 were located in the pineal region. One child presented with intracranial leptomeningeal disseminated disease. The tumors were mostly undifferentiated although 22 had some evidence of differentiation along one or more neuroepithelial lines. Five children had a biopsy, 24 had subtotal resection, and 7 had gross total resection. Twenty-six children had adjuvant radiotherapy and 13 had chemotherapy. At last follow-up 30 patients were dead and 6 were alive. The median survival was 23 months and the 2, 3, and 5 year survivals were 50%, 34%, and 18% respectively. All of the survivors received craniospinal radiation and 4 received chemotherapy. There was a statistically significantly worse survival in young children. There was a trend to better survival in children treated since 1984, and in children undergoing gross total resection. Because of the extremely poor survival, we recommended that all children undergo gross total resection followed by chemotherapy. For children older than 3 years of age craniospinal radiation should also be given.

Analysis of glial fibrillary acidic protein gene methylation in human malignant gliomas.

Glial fibrillary acidic protein (GFAP) is an intermediate filament specifically expressed in glial cells which helps to maintain and stabilize the glial cytoskeleton. Interestingly, with increasing astrocytic anaplasia, there is typically progressive loss of GFAP expression. In in vitro model systems, most permanent glioma cell lines are GFAP-negative. To determine the mechanism by which the transcription of the GFAP gene may be repressed in glioma cell lines, we initially performed a Southern analysis on a panel of human malignant glioma cell lines using a human cDNA probe for GFAP. By this method, no large rearrangements or deletions of the GFAP gene were found. Postulating that a change in methylation status of the GFAP gene could conceivably alter its expression in glioma cell lines, we studied the methylation state of the GFAP gene in the same glioma cell lines using a methyl-sensitive restriction enzyme digest of tumour and control DNA. Our analysis revealed that the GFAP gene was hypermethylated in 2/2 GFAP- negative glioma cell lines but not in 4/4 GFAP-positive glioma cell lines. To determine if methylation of CpG islands contained within the GFAP promoter could repress GFAP transcription, we designed deletional constructs from a 6 kb fragment of the mouse GFAP promoter, methylated them using Msp I- and Hpa II-methylases, and tested their activity in a standard CAT assay. Our data suggest that methylation of a 2 kb segment of the mouse GFAP promoter is sufficient to inactivate GFAP transcription. Our results further imply that methylation-mediated repression of GFAP transcription may be one candidate mechanism to account for decreased GFAP expression in certain human malignant glioma cell lines.

Transfection of an invasive human astrocytoma cell line with a TIMP-1 cDNA: modulation of astrocytoma invasive potential.

Malignant astrocytomas are highly invasive tumors which infiltrate diffusely into regions of normal brain. The degradation of the extracellular matrix (ECM) by matrix metalloproteinases is thought to be one of the most important steps in the process of tumor invasion. However, the activity of most matrix metalloproteinases (MMPs) can be modulated by simultaneously secreted inhibitors (tissue inhibitors of metalloproteinases, TIMPs). We have previously shown that an imbalance between the levels of MMPs and TIMPs may be essential in the determination of the invasiveness of certain human malignant astrocytoma cell lines. To determine if the up-regulation of TIMP genes and gene products could modulate the invasiveness of human malignant astrocytoma cells, in the present study we have transfected a highly invasive astrocytoma cell line, SF-188, with an expression vector carrying a full-length TIMP-1 cDNA. The parental SF-188 astrocytoma cell line overexpresses the 72-kDa and 92-kDa type IV collagenases with little expression of TIMPs-1 and -2. Following transfection with TIMP-1, SF-188 astrocytoma clones expressed the 0.9 kb TIMP-1 message by northern analysis, and produced a 21 kDa metalloproteinase inhibitor by reverse zymography. The stable TIMP-1 SF-188 transformants demonstrated morphological changes and diminished growth rates in soft agar when compared to controls. The invasion of successfully TIMP-1 transfected astrocytoma cells across matrigel-coated filters was significantly decreased over controls. These results suggest that upregulation of TIMP-1 expression in SF-188 astrocytoma cells has decreased their in vitro invasive potential.