Evaluation of O6-methylguanine-DNA methyltransferase (MGMT) expression is important for antiglioma therapy as many clinical trials have demonstrated that promoter hypermethylation and low level expression of MGMT are associated with an enhanced response to alkylating agents. However, here we report that the current strategies used to evaluate MGMT status in gliomas are unreliable. We observed discordance in the MGMT expression status when immunohistochemical evaluation and polymerase chain reaction-based methylation assessments were used: 73% of gliomas with methylated MGMT promoter had substantial numbers of MGMT-immunopositive tumor cells. Furthermore, when MGMT expression was tested in tumor homogenates using reverse transcription-polymerase chain reaction, 43% of tumors were found positive, in comparison to only 24%, when histologic samples were assayed immunohistochemically. To explain these inconsistencies we undertook a detailed immunohistochemical evaluation of tumor samples and found that some gliomas demonstrated remarkably high expression of MGMT in the entire tumor whereas others contained only a small immunopositive area. Additionally, we found that gliomas contained various types of non-neoplastic cells expressing MGMT, including lymphocytes, vascular endothelial cells, and macrophages/microglias, which contribute to overall MGMT expression detected in tumor homogenates, and thus result in overestimation of tumor MGMT expression. Therefore, to correctly establish MGMT expression in the tumor, which could be informative of glioma sensitivity to alkylating agents, exclusion of non-neoplastic brain components from analysis is required.

Glycosylation is one of the most common posttranslational modifications and changes in oligosaccharide structures are associated with many human diseases including a number of cancers. Thus, discovering aberrant glycosylation patterns that serve as markers for brain tumor progression and metastasis represents an attractive strategy to improve clinicopathologic diagnosis and to provide aids to the development of novel therapies. To identify glioblastoma (GBM) cells expressing glycoproteins that contain high levels of the bisecting N-acetylglucosamine (GlcNAc) structures, lectin histochemistry was carried out using erythroagglutinating phytohemagglutinin. Although GBM frequently expressed the bisecting GlcNAc, the lectin reactivity varied among tumor regions within individual specimens. Since detailed histopathologic analysis revealed that oligosaccharides with bisecting GlcNAc structures were preferably expressed in tumor regions with low KI67 immunopositivity, immunodetection of the bisecting GlcNAc could be useful to indicate less proliferative regions in human GBM. Our study highlights the potential use of lectin histochemistry to develop new methods for diagnosis that would improve future antiglioma therapy.

Overexpression of transcription factors is often observed in malignant gliomas and the roles of such factors in tumor malignancy have been widely investigated. However, the contribution of glioma-related transcription factors to tumor initiation are largely unclear, as the majority of brain tumor studies have employed glioma-derived cell lines, which recapitulate only a subset of late-stage human tumors. In contrast to tumor-derived cell lines, which harbor numerous undefined genetic elements, normal human cells allow us to generate experimental models of human cancers by direct manipulation with defined genetic elements. Here we demonstrate that the introduction of GLI1 is sufficient for immortalized human astrocytes to be transformed whereas FOXM1 fails to induce oncogenic transformation, suggesting that GLI1, but not FOXM1, directly contributes to gliomagenesis despite both transcription factors being overexpressed in malignant gliomas. While GLI1-transformed astrocytes have the ability to form tumors in nude mice, the oncogenic potential of GLI1 is subtle compared to the active form of H-RAS. To obtain insights into the mechanisms underlying less malignant features of GLI1-transformed astrocytes, expression of inhibitors of cyclin-dependent kinases in xenografts were analyzed by immunohistochemistry. We found that p27KIP1-positive cells were frequently observed in xenografts derived from GLI1-transformed astrocytes (GLI1-xenografts) compared to those from RAS-transformed cells (RAS-xenografts): more than 60% of cells were positive for p27KIP1-staining in GLI1-xenografts whereas only ∼20% in RAS-xenografts. We also made an interesting observation that CD133-positive cells were frequently detected in GLI1-xenografts whereas they were rarely seen in RAS-xenografts. The biological relevance and regulatory mechanisms of CD133 expression in human glioma are not completely understood although the use of CD133 as a marker to sort and enrich for brain tumor stem cells is common. Since GLI1 failed to induce CD133 expression in cultured astrocytes, it is possible that additional factors present in vivo and absent in culture might collaborate with GLI1 to modulate CD133 expression and that tumor microenvironment might contribute to maintenance of brain tumor stem cells. We propose that GLI1-tranformed astrocytes may prove to be useful tools as we strive to understand the molecular mechanisms underlying CD133 expression in gliomas as well as maintenance of brain tumor stem cells. Citation Information: Cancer Res 2009;69(23 Suppl):C1.