States and transitions in the initiation and maintenance of melanoma

PROJECT SUMMARY The presence of combinations of oncogenic gene mutations is characteristic of many types of cancer. When such mutations are genetically engineered into mice, the resulting cancers have proved to be valuable models of human disease. Yet the fact that only a fraction of the appropriately engineered cells in such models typically give rise to tumors suggests that additional random steps—other than mutation—are required for tumor initiation. In mouse models of melanoma, there is growing evidence that some of those steps are non- genetic involving, for example, stochastic switching of cell fate—a process that has been documented in melanoma cell lines, and is thought to account for the high degree of cellular heterogeneity observed in melanoma. Another type of stochastic, non-genetic change can occur as a result of collective processes, whereby random fluctuations in cell number, density, or spatial organization allow populations of cells to permanently escape growth control. Here we analyze a series of mouse melanoma models that vary in the number of engineered mutations they contain and numbers and types of malignant and benign lesions they produce, seeking to elucidate the non-genetic mechanisms underlying cancer initiation. We start by using single cell transcriptomics and novel computational methods to rigorously define the cell states that are available to melanoma cells in different genotypes, determine when they arise, infer their transition probabilities, and examine which of these states are present in human melanoma and how they are spatially organized. Next, we use molecular barcoding approaches to directly track cell state transitions, identify the potential molecular regulators of these transitions, examine potential symbiotic relationships between cell states, and examine how the innate immune system influences tumor initiation and cell state transitions. To test whether cancer is initiated by collective escape from growth control, we will assess experimentally whether, just by altering cellular circumstances, cells committed to become cancer can be reverted to normal, and cells destined to behave normally can be driven to become cancer. The work described here has the potential to change established views of melanoma initiation, identify new approaches for early detection, and lead to the identification of new therapeutic strategies for melanoma prevention.