It is common to hear scientists equate targeting cancer to playing whack-a-mole: hitting one “mole” (or pro-tumorigenic signal) will lead to the emergence of another mole, then another, and so on. This analogy is far too simple and disregards the elegant evolution observed in cancer. In this respect, I propose that cancer is more comparable to a pin art board. At one moment the pins are in the shape of a hand. If asked to describe the pin art board, all would agree that fingers and a palm are critical components. In the next moment the hand is replaced with a face defined by eyes and a mouth. Although these two snapshots seem quite different, there are underlying mechanics that can explain the fluidity that permits the hand to morph to a face, and those mechanics are shared across all pin art boards. In a similar way, cancer is quite fluid. At one moment the tumor is addicted to a signaling axis that can be therapeutically targeted. Upon blockade of that addiction, the tumor is able to elegantly morph into a new state with an entirely new set of defining features. This “new tumor state” is not a random “mole” that sporadically pops up in an unpredictable manner; rather, the tumor can predictably hijack mechanisms that permits fluid state changes. Instead of focusing on the mole, or better yet the shape of the pin art board, here we try to understand the mechanics and drivers of what defines the observed tumor phenotype. What allows tumor cells to overcome toxic insults, and adapt to novel and harsh environments?

To this end, tumor initiation, therapy resistance, and metastasis are highly dependent on the ability for cells to adapt and survive, and thus represent processes that select for the most adaptable, aggressive cell type. The following thesis aimed to identify and characterize tumor cells with the intrinsic capacity to thrive during those processes. Particular focus was given to tumor cells with stem cell properties, as these are characteristically able to respond to stresses and regenerate tumors accordingly. Using pancreatic cancer as a model system, we identified a population of tumor cells, functionally marked by the stem cell fate determinant Musashi, with cancer stem cell characteristics: Musashi+ cells were enriched for tumor propagating capacity, highly resistant to cytotoxic therapies, and preferentially survived in circulation. Further, we carried out unbiased screens to understand the landscape that defined these pancreatic cancer stem cells. This led to the discovery of targetable signals required for maintenance of a stem cell state, and thus represents an exciting new therapeutic approach designed to collapse mechanisms that promote an adaptable, aggressive cell phenotype. Finally, I highlight the power of using intravital imaging to track tumor cell behaviors in vivo, and apply this tool to better understand intrinsic and extrinsic dependencies.