Diffuse brain invasion is a major reason for poor prognosis of glioma patients. The molecular mechanisms underlying infiltration are different from those of other cancer types. To detect genes associated with glioma invasion, highly migratory clones were selected from U373MG glioma cells and from primary glioblastoma cells, and the gene expression pattern of these "fast" cells was compared with that of the original ("slow") cells using oligonucleotide microarrays comprising 12,625 genes. A total of 28 genes were differently expressed in both primary and established cell populations, including 19 genes that were upregulated and 9 that were downregulated in fast cells. Most of these genes have not been linked to glioma invasion so far. Specifically, differentially expressed genes included those encoding extracellular matrix components (COL16A1, DPT), proteases (CATD, PRSS11), cytokines (MDK, IL8), transport proteins (SLC1A3, ATP10B), cytoskeleton constituents (ACTA2, ACTSG, NEFL), DNA repair enzymes (WRN, ADPRTL2), and G-protein signaling components (GNA12, RGS3, RGS4). RGS3 and RGS4, which are homologs of the Drosophila glia gene loco, were further functionally analyzed. U373MG glioma cell clones overexpressing RGS3 or RGS4 showed an increase of both adhesion and migration. These findings expand the spectrum of possible molecular pathways underlying the invasion of neoplastic astrocytes. Specifically, they suggest that RGS proteins and G-protein-mediated signal transduction are evolutionary conserved functional players.

Meningioma represents the most common intracranial tumor, but well-characterized cell lines derived from benign meningiomas are not available. A major reason for the lack of benign tumor cell lines is senescence of nonmalignant cells in vitro, while malignant cells are often immortal. We have developed a meningioma cell line by retrovirally transducing primary cells derived from a human WHO grade I meningothelial meningioma with the human telomerase reverse transcriptase (hTERT) gene, which enables bypassing cellular senescence. Five clones have been cultured for more than 21 months so far, while corresponding nontransfected cells ceased proliferation within 3 months. Quantitative RT-PCR and a telomeric repeat amplification protocol (TRAP) assay revealed high hTERT mRNA levels and high telomerase activity in all transduced populations, while nontransduced cells were negative. The average telomere size of transduced cells was considerably longer than that of parental cells and the biopsy specimen. One clone, designated Ben-Men-1, was characterized in more detail, and exhibited typical cytological, immunocytochemical, ultrastructural and genetical features of meningioma, including whorl formation, expression of epithelial membrane antigen, desmosomes and interdigitating cell processes, as well as À22q. Following subdural transplantation into nude mice, tumor tissue with typical histological features of meningothelial meningioma was found. We conclude that Ben-Men-1 represents an immortalized yet differentiated cell line useful for biological and therapeutical studies on meningioma.

The beta2 subunit of Na,K-ATPase, initially described as adhesion molecule on glia (AMOG), has been shown to mediate neurone-astrocyte adhesion as well as neural cell migration in vitro. We have investigated the expression of AMOG/beta2 in human gliomas and its effect on glioma cell adhesion and migration. Compared to normal astrocytes of human brain, AMOG/beta2 expression levels of neoplastic astrocytes were down-regulated in biopsy specimens and inversely related to the grade of malignancy. One rat and four human glioma cell lines showed complete loss of AMOG. To investigate the function of AMOG/beta2, its expression was re-established by transfecting an expression plasmid into AMOG/beta2-negative C6 rat glioma cells. In vitro assays revealed increased adhesion and decreased migration on matrigel of AMOG/beta2-positive cells as compared to their AMOG/beta2-negative counterparts. We conclude that increasing loss of AMOG/beta2 during malignant progression parallels and may underlie the extensive invasion pattern of malignant gliomas.

Identification of genes mediating glioma invasion promotes the understanding of glia motility and might result in biologically based therapeutic approaches. Most experimental studies have been performed in vitro, although glial cells typically undergo marked phenotypic change following placement into cell culture. To evaluate migration mechanisms operating in vitro versus in vivo, we used C6 rat glioblastoma cells for selecting highly migratory cells in a monolayer migration assay as well as in brains of nude mice, and analyzed in each paradigm the expression profiles of these "fast" cells versus those of the original "slow" cells using oligonucleotide microarrays comprising 8832 genes. In vitro, 516 (10.6%) of 4848 expressed genes were regulated (i.e., differentially expressed in fast versus slow cells); 916 genes were expressed only in vitro, including 142 (15.5%) regulated genes. In vivo, 245 (6.1%) of 4044 expressed genes were regulated; 112 genes were expressed only in vivo, including 25 (22.3%) regulated genes, none of them having a known relation to glioma invasion. Of 730 regulated genes, only 31 (4.2%) were regulated in parallel in vitro and in vivo, most of them having a known relation to (glioma) invasion. Our data provide new molecular entry points for identifying glioma invasion genes operating exclusively in the brain. They further suggest that genes underlying glia cell motility are strikingly different in vitro and in vivo.

L1 is an adhesion molecule of the immunoglobulin superfamily expressed by several types of cancer, including gliomas. It has been shown that L1 can act as chemoattractant to glioma cells, while the effects of L1 expressed by glioma cells themselves are unknown to date. We established a C6 rat glioma clone, conditionally expressing murine L1 under control of a tetracycline responsive promoter. In vitro experiments revealed increased adhesion on matrigel as well as increased intercellular adhesion in the presence of L1, whereas no L1-dependent effects on proliferation or migration on either matrigel or myelin were observed. In vivo experiments using transplantation into nude mouse striatum, where L1 expression by glioma cells was regulated by tetracycline via drinking water, did not show effects of L1 on tumor size or brain invasion. Our data suggest that L1 expressed on the surface of glioma cells increases cell-matrix and intercellular adhesion, but has no apparent effects on proliferation and invasion.