Glioblastoma multiforme (GBM) is amongst the most lethal of all cancers. GBM consist of a heterogeneous population of tumor cells amongst which a tumor initiating and treatment-resistant subpopulation, here termed GBM stem cells (GSC), have been identified as primary therapeutic targets. Here, we describe a high-throughput small molecule screening approach that enables the identification and characterization of chemical compounds that are effective against GSC. The paradigm uses a tissue culture model to enrich for GSC derived from human GBM resections and combines a phenotype-based screen with gene target-specific screens for compound identification. We used 31,624 small molecules from seven chemical libraries that we characterized and ranked based on their effect on a panel of GSC-enriched cultures as well as their effect on the expression of a module of genes whose expression negatively correlates with clinical outcome: MELK, ASPM, TOP2A and FOXM1b. Of the 11 compounds meeting criteria for exerting differential effects across cell types used, 4 compounds demonstrated selectivity by inhibiting multiple GSC-enriched cultures compared to non-enriched cultures: Emetine, N-Arachidonoyldopamine (NADA), N-Oleoyldopamine (OLDA), and N-Palmitoyldopamine (PALDA). ChemBridge compounds #5560509 and #5256360 inhibited the expression of the 4 mitotic module genes. OLDA, Emetine, and compounds #5560509 and #5256360 were chosen for more detailed study and inhibited GSC in self-renewal assays in vitro and in a xenograft model in vivo. These studies demonstrate that our screening strategy provides potential candidates as well as a blueprint for lead compound identification in larger scale screens or screens involving other cancer types.

No abstract

No abstract

ABSTRACT. This paper deals with some important decisions that ought to be made when building a mixture-of-experts model (MEM). Such decisions are related to aspects such as the clustering method and the gating functions used in the model. Depending on how these decisions are made, different mixtures might be formed, yielding different results. In the present study, we investigate the way such decisions affect the performance of MEM's, when using statistical models to regression problem. The famous Boston housing data is used as illustration for the technique.