To test our hypothesis, we have two specific aims described below:
Specific Aim 1. Characterize the invasive cells in vitro and in vivo. The invasive versus noninvasive populations in GBM stem cells will be separated and collected from the matrigel invasion assay in vitro. The ability to form invasive tumors in vivo of these two distinct populations will be analyzed by injecting the cells into host mice brains. Meanwhile, immunostaining and western blotting for can-cer stem cell markers, including but not limited to CD133, SSEA-1 (CD15), A2B5, CD171 (L1CAM), CD49f (integrin-alpha6), CD44 and EGFR, will be performed to compare these two populations in vitro. Realtime PCR will be performed if an antibody is not available. Soluble motility factors (such as SDF-1/CXCR4) that function as autocrine or paracrine signaling will also be analyzed by immunostaining, western blotting and/or realtime PCR. Furthermore, RNAs isolated from these two populations will be subject to transcriptome analysis to get a full list of differentially expressed genes. We will characterize the properties of invading cells in vivo using the syngeneic intracranial tumor model. As a perivascular niche for GBM stem cells has been reported, we will visualize the interactions between donor cells and the host vascular system by labeling both cell populations with fluorescent proteins. To label the blood vessels in the host brain, a Tie2-Cre mouse with constitutive Cre expression in endothelial cells of the adult vasculature will be bred to a Rosa26-YFP reporter strain, which strongly expresses the yellow fluorescent protein, following Cremediated recombination. GBM stem cells will be labeled with lentiviral expressed RFP under CMV promoter. Flow cytometry sorting will be performed to enrich for RFP positive cells before intracranial injection. A time course analysis will be performed by collecting brain samples every week after injection for 5 weeks or until the mice show obvious signs of tumor burden. To determine the pattern of tumor cell migration, RFP positive tumor cells and YFP positive host endothelial cells will be monitored by fluorescence microscopy, both on fixed brain sections and cultured live brain slices. For brain slice culture, the host brains will be dissected manually and cut into 150-micrometer coronal slices using a vibratome. Slices will be cultured on Millicell-CM culture plate inserts (Millipore) in six-well glass bottom plates, and will be maintained in the neural stem cell culture medium. Serial images will be acquired every 15 minutes with a fluorescence microscope. Immunostaining for cancer stem cell markers and soluble motility factors described above will be performed to assess the properties of invading cells at the diffusive tumor margin. In situ hybridization will be performed alternatively if an antibody is not available. Laser capture microdissection will be performed to collect cells in the invasive rim and the tumor core from brain sections. RNA samples will be isolated from both cell populations and subject to transcriptome analysis to examine differences in the expression profile between invading and noninvading cells. Together, results obtained from these studies will provide useful information about the invasive cells in GBM and potentially identify useful biomarkers for cells that invade. These data will also be used to integrate with the RNAi screen results obtained in specific aim 2.
Specific Aim 2. Identify functional regulators of GBM invasion. We will attempt to identify novel regulators of GBM invasion using RNAi screening as a systematic functional genetic approach. Though it is the best to perform such a screen in the invading cells isolated from GBM, no markers are yet available for isolating these cells from the tumor directly. Therefore, we will use the GBM stem cell lines established from human patients (provided by Dr. Alfredo Quinones-Hinojosa, Johns Hopkins University) for the screen, as they contain an invasive subpopulation that migrates through the matrigel invasion chamber, and can form highly invasive GBM in the intracranial xenograft models. The advantage of using human GBM stem cells is that targets more relevant to human disease may be identified from the screen. Then candidate targets can be validated, and potential mechanisms can be dissected in the mouse system. The matrigel invasion assay will be used to enrich for stem cells with invasive properties. A genome wide pooled human shRNA library (provided by Dr. Ji Luo, National Cancer Institute) will then be used to knock down gene expression in the invasive stem cells, with puromycin selection to eliminate uninfected cells. Following knock-down, we will assay for cell mobility with the matrigel invasion assay again. Cells that remain on top of the matrigel membrane and that migrate through the membrane will both be collected, propagated and sampled for bar code array analysis to identify candidate shRNAs that inhibit cell invasion. Corresponding genes will be subject to bioinformatic analysis and those relevant to GBM based on gene expression will be individually tested by a set of different shRNAs to minimize off target effects. The primary screen hits will also be repeated in a second cell line to ensure the reproducibility of the phenotype. Candidate shRNAs resulting from the screen will be further functionally validated in the intracranial xenograft model. The invasive potential of cells expressing candidate shRNAs will be evaluated in host mice. Cells expressing scrambled shRNA will be used as controls. If a small molecule inhibitor is available for a particular factor, it will be tested in both the intracranial xenograft model and the de novo GEM-GBM mouse model. We will further perform cross-species analyses for successfully validated shRNAs in the mouse system. For candidate hits that are conserved in mice, we will use the syngeneic intracranial GBM model described in specific aim 1, as well as the de novo GEM-GBM models, for further dissection of underlying mechanisms and preclinical evaluation.
