Abstract Malignant cancers evolve over time with multiple contributing oncogenic mutations, and contain both tumor and niche cells. To understand when mutations contribute to malignant progression and how tumor and niche cell interact, we use a novel mouse genetic system termed MADM (Mosaic Analysis with Double Markers). MADM enables the tracking of tumorigenic cells from early stages with single cell resolution in vivo. Starting with a mouse heterozygous for a tumor suppressor gene (TSG), MADM can generate sporadic TSG-null cells via Cre-loxP mediated inter-chromosomal mitotic recombination events. Further, MADM unequivocally labels mutant cells with GFP, enabling lineage tracing of mutant cells with clear distinction from neighboring normal cells. Following cell division generating a mutant GFP+ cell, the sibling wildtype (WT) cell (RFP+) can also be assessed as an in situ internal control. We have applied MADM to model the Sonic Hedgehog (SHH)-subtype of medulloblastoma, the most common pediatric brain tumor. In the developing cerebellum, unipotent granule neuron progenitors (GNPs) proliferate at the cerebellar surface in response to active Sonic hedgehog (SHH) signaling. Somatic mutation of the SHH receptor, Patched (Ptc), can lead to hyperactivation of the pathway, GNP overproliferation, and eventual malignancy. In addition to Ptc mutation, it has been reported that patients with medulloblastoma have a much worse prognosis when the tumor suppressor TRP53 (p53) is mutated. However, the timing of p53 contribution to tumorigenesis is still unclear. Here we used MADM to inactivate p53 in sporadic GNP cells in a Ptc heterozygous mouse model. To pinpoint the timing of p53 contribution to tumor progression, we performed a time course analysis of green-to-red cell number ratio (G/R ratio or mutant/wildtype). Our preliminary data suggest that the loss of p53 does not contribute to tumor progression until mid-stages of tumor progression since G/R ratio remains 1 in early lesions induced by the Ptc mutation. For spatial organization of tumor and niche cells, we took advantage of the lineage tracing ability of MADM since all tumor cells should be green while neighboring normal cells should be non-labeled. Because the loss of heterzygousity is induced in unipotent GNPs, we expect the GFP+ tumor cells to maintain this identity. Surprisingly we find GFP+ cells in tumors consist of two cell types, GNPs and astrocytes, suggesting the trans-differentiation of some mutant GNPs into astrocytes. To confirm the human relevance of our findings, we performed double FISH-IF analysis of human patient samples for PTCH and our preliminary data indicated that GFAP+ astrocytes within tumors carried similar genomic aberrations to tumor GNPs, implicating the lineage relationship between these two cell types. Finally, in vitro co-culture experiment provided initial evidence that tumor GNP-derived astrocytes provide support to tumor GNPs, hinting at intricate cellular interactions in medulloblastoma. In summary, our findings indicated that MADM provides great temporal and spatial resolution for cancer research. Our preliminary data suggest that tumor progression is not a linear process and distinct tumor suppressor genes act at different stages, and that tumor-niche interactions could be more complex than previously thought. Citation Format: Brit Ventura, Maojin Yao, Ying Jiang, Kelsey Wahl, Fausto Rodriguez, Charles G. Eberhart, Hui Zong. Fine temporal and spatial dissection of medulloblastoma progression with MADM, a mouse genetic mosaic model. [abstract]. In: Proceedings of the Third AACR International Conference on Frontiers in Basic Cancer Research; Sep 18-22, 2013; National Harbor, MD. Philadelphia (PA): AACR; Cancer Res 2013;73(19 Suppl):Abstract nr PR09.