BACKGROUND: We used human fetal neural stem cells dissected from the cerebral cortex, thalamus, cerebellum and brainstem, then propagated as neurospheres, to model a range of brain tumors. Various combinations of oncogenic elements characteristic of medulloblastoma and glioblastoma were introduced into stem cells derived from these different sites. Transformation of cells was characterized in vitro and in vivo, and MYC driven medulloblastoma were used for preclinical testing of glutaminase inhibitors. METHODS: We transduced cerebellar derived human neural stem cells with lentivirus coding for MYC, dominant negative R248W p53, constitutively active AKT, and hTERT in several combinations. Growth of transformed cells and their response to therapy was analyzed in vitro and in vivo using standard techniques. RESULTS: MYC alone was sufficient to transform cerebellar progenitor cells into embryonal tumors with a predominantly neuronal phenotype, but with a long latency period and decreased penetrance suggesting secondary genetic changes were required. The R248Wp53/MYC/AKT/hTERT combination transformed cerebellar stem cells into more aggressive orthotopic xenografts with large cell/anaplastic histology and spinal metastases. Microarray expression analysis suggested that these tumors represent a model of high MYC group 3 medulloblastoma. MYC expression positively correlated with increased expression of glutaminase, and the glutaminase inhibitor Acivicin decreased cell proliferation more strongly in MYC-containing neurospheres compared to those without MYC. Acivicin treatment of MYC expressing cells also led to a much greater increase in apoptosis compared to MYC-negative cells. Introduction of other packages of oncogenic elements into cortical and thalamic neurospheres resulted in PNET-like tumors, including variants resembling medulloepithelioma, as well as glioblastoma. CONCLUSIONS: Human cerebellar stem cells can be transformed by c-MYC and cooperating oncogenes, resulting in a models that recapitulates group 3 medulloblastoma histologically and clinically. These cells and classical high-MYC medulloblastoma cell lines are sensitive to glutamine anti-metabolites, while non-MYC driven medulloblastoma and SV40 immortalized human cerebellar stem cells are resistant. Our group 3 human neural stem cell medulloblastoma model data suggest that glutamine metabolism may be a therapeutic target in MYC-driven medulloblastoma, and that glutamine anti-metabolites may be useful clinical agents. Neurospheres from other CNS sites can be transformed with other oncogenes into models of PNET and glioblastoma. SECONDARY CATEGORY: Tumor Biology.