With an estimated 190,000 new diagnoses in 2020, Prostate Cancer (PC) is the second most frequent tumor diagnosed in men and the second leading cause of male cancer deaths in America. Approximately one in three men over the age of 50 shows histological evidence of this tumor, however only one in ten will be diagnosed with clinically significant PC. Therapeutic options for localized PC are effective, however, metastatic PC is a yet incurable disease because standard of care anti-androgen therapy invariably results in incurable disease relapse. PTEN-TP53 loss is a most significant event of human lethal metastatic PC as summarized by the PCF/SU2C International Prostate Cancer Dream Team (Armenia et al., 2018) and confirmed by two decades of functional PC modeling in mouse. The central goal of this project is to understand what causes PTEN-mutant indolent prostate cancer (PC) to metastasize to different parts of the body and kill the patient. While the PCF/SU2C report significant association of PTEN loss with loss of TP53, the definition of further genes that are significantly co-mutated with PTEN has remained a problem. Importantly, such genes could point to pathways and principles that drive metastasis and/ or therapy resistance. We aim to solve this problem by combining single cell analysis of patient metastatic rapid autopsy samples with functional genetics and 3D whole organ imaging of lethal metastasis in mouse. We use a highly flexible genetically engineered mouse (GEM) model of lethal, endogenous metastatic prostate cancer, termed RapidCaP. Spontaneous progression to lethal metastasis is seen in ~70% of Pten/Trp53-mutant RapidCaP, suggesting a critical stochastic pioneer event that switches a Pten/ Trp53 null tumor cell from indolence to metastatic escape, similar to a critical pioneering event that causes relapse after castration therapy. However, classical approaches can only capture and reveal the time and physical nature of these pioneering events when already thousands or millions of cells are involved. In Aim 1, we reveal, isolate, and analyze single pioneer cells in space and time. We combine RapidCaP with serial two photon tomography (STPT), which allows us to image whole organs at single cell resolution and construct a 3D map of the time course of escape from indolence and of metastatic relapse after therapy. This guides our isolation of cell types that then serve as validated proxies for pioneer cells of escape or relapse. Through Aim 2, we solve the PTEN co-mutation impasse and define genes co-deleted with PTEN through single cell whole genome analysis of human rapid autopsy samples. We prioritize candidates based on comprehensive functional molecular probing of already isolated pioneer cells in vitro. Through Aim 3, we develop two new modeling platforms for more flexible validation of candidate genes and principles behind escape from indolence and relapse from castration therapy.