Levi A. Garraway - US grants

DESCRIPTION (provided by applicant): It is clear that inactivation of PTEN is a major contributor to the evolution of hormone-refractory prostate cancer. The function of PTEN as a tumor suppressor is most closely linked to its activity as a lipid phosphatase (Maehama and Dixon, 1998) and hence its ability to attenuate signaling through the phosphoinositide 3-kinase pathway. In the initial funding period our work focused on the role of PTEN as a cell-cycle regulator. Here,we showed that the PTEN lipid phosphatase activity and regulation of Akt was essential for PTEN to arrest cells in G1 (Ramaswamy et al., 1999). Elegant genetic studies placing PTEN in the dauer pathway in C.elegans led us to ask whether human FOXO homologues of daf-16, might play a role in PTEN-mediate tumor suppression. Indeed, we found that Foxol is constitutively inactivated in PTEN null cells and its reconstitution is sufficient to reverse the transformed phenotype of PTEN-null cancer cell lines (Nakamura et al., 2000). Surprisingly, we found that this Foxol activity did not require binding to an insulin-response element and instead was associated with the ability of FOX01 to inhibit cyclin D1 and D2 transcription (Ramaswamy et al., 2002). Preliminary data strongly suggest that FOX01 can act as a transcriptional repressor through interactions with the proto-oncogene SKI and that this activity is linked to its ability to induce a cell-cycle arrest and suppress tumor growth. Based on this data we propose in specific Aim 1:1. To test the hypothesis that a repressor complex mediates the cell-cycle regulatory properties of FOXO1- During the original grant period we were the first group to demonstrate that PTEN is phosphorylated on residues within the C-terminal 'tail' and that these phosphorylation events appear to regulate PTEN stability. Moreover, we showed that the unphosphorylated form of PTEN is found in a more open conformation, associates with a high molecular weight complex, and is more active in biological assays. Based on this data we proposed a model whereby the majority of cellular PTEN is phosphorylated and in a relatively inactive state, upon dephosphorylation PTEN can then enter a so-called PTEN-associated complex (PAC) (Vazquez et al., 2001). In this context PTEN is more active yet subjected to degradation. Data from other labs have been entirely consistent with this model, yet an understanding of the mechanism through which phosphorylation regulates PTEN, the localization of the unphosphorylated forms of PTEN, and the nature of the PAC remain obscure. Based on this data we propose in specific aims 2 and 3: 2. To generate antibody reagents that selectively recognizes unphosphorylated PTEN. 3. To purify and identify proteins in the PTEN associated complex (PAC).

DESCRIPTION (provided by applicant): Melanona represents an important cause of cancer-related mortality in the Western world, and its incidence has increased steadily over the past 3 decades. Advanced melanoma remains largely refractory to most therapeutic approaches; in part, this may reflect the considerable genetic complexity underlying melanoma and most other solid tumors. To this end, a main goal of cancer genomics is to characterize genetic alterations at a large scale within human tumors, in hopes of deriving molecular classification schema that might aid targeted therapeutic interdiction. High-density SNP microarrays, which simultaneously interrogate more than 100,000 SNP alleles, have recently been adapted for high-resolution cancer genome analysis. Prior work by the Sellers/Meyerson group and others showed that SNP arrays enabled detailed characterization of chromosomal gains, losses, and loss of heterozygosity (LOH) regions, even in the absence of matched normal samples. In addition, preliminary studies on the NCI60 panel of cancer cell lines demonstrated that combined analysis of SNP array and gene expression data, followed by functional validation, represented a powerful means of tumor classification and cancer gene identification. We are therefore applying an integrated, functional genomic approach to the study of metastatic melanoma. Specifically, high-density SNP array and comprehensive gene expression data will be collected on a large set of patient-derived, short-term melanoma cultures. Computational algorithms will then be employed to define and order the meaningful deletions, amplifications, and LOH regions across the sample set. Next, gene expression data will be combined with chromosomal information to identify candidate cancer genes within relevant genomic lesions. Finally, selected candidate oncogenes will be validated functionally by a lentiviral shRNA approach. These approaches should provide robust grounds for melanoma classification and gene discovery, thereby offering new options for future therapeutic interdiction.

NIH Roadmap Initiative tag; functional /structural genomics; neoplasm /cancer genetics; neoplasm /cancer therapy; therapy design /development

[unreadable] DESCRIPTION (provided by applicant): ' Cancer represents a disease of the genome; each tumor harbors a distinct set of mutations that activate oncogenes and inactivate tumor suppressor genes. In the era of targeted therapeutics, it is expected that cancer treatment decisions will increasingly be made based on tumor genetic composition as opposed to tissue of origin. However, most molecular diagnostics that detect cancer gene mutations are expensive, informative for only a single genetic locus, and adversely affected by degraded or stromally admixed genomic DNA. Thus, despite the promise of somatic cancer genetics, at the present time it remains impractical to identify critical oncogene mutations on a large scale and in a manner compatible with routine clinical use. To address these limitations, this application aims to adapt a high-throughput, mass spectrometry- based genotyping technology to detect somatic mutations in a large panel of cancer genes.' In the R21 phase, a platform based on SequenomTM iPLEX genotyping will be developed that interrogates over 600 point mutations (or small insertions/deletions) across 50 oncogenes and selected tumor genes. This platform will also be optimized for cancer gene mutation detection in genomic DNA from paraffin-embedded tumor tissue. In the R33 phase, test the feasibility of this mutation detection approach will be demonstrated in a study of a large and diverse tumor collection. Here, high-throughput oncogene mutation detection will be performed on nearly 2,700 frozen and paraffin-embedded tumors spanning many lineages, including several that have not undergone prior genomic characterization. The ability to perform high-throughput mutation detection in clinical tumor samples will enable unprecedented molecular analyses applicable to molecular epidemiology and translational oncology, including patient stratification for targeted cancer therapeutic trials. These studies therefore offer immense potential to benefit investigators and patients alike on the path to rational cancer therapeutics. [unreadable] [unreadable] [unreadable]

DESCRIPTION (provided by applicant): Cancer diagnosis and treatment decisions have historically been based on anatomic sites of origin and spread;however, the emerging paradigm incorporates key genetic attributes of a given tumor to predict clinical behavior and specify the optimal use of targeted therapeutics. Ultimately, the delivery of personalized cancer medicine will require systematic characterization of all therapeutically informative tumor genomic alterations in the clinical and translational arena. Previously, we developed and deployed OncoMap, a mass spectrometric genotyping-based platform that enables high-throughput profiling of hundreds of known mutations across dozens of cancer genes. This platform performs well and has launched a robust translational oncology effort. However, the mass spectrometric genotyping technology remains limited in scope, assay sensitivity, and the type of genomic alteration that can be identified. Recently, it has become possible to render multi-faceted tumor characterization both technologically feasible and economically accessible through massively parallel sequencing (MPS) technology. Thus, the goal of this application is to migrate the OncoMap approach to an MPS platform (Illumina), empowered by innovations such as solution-phase exon capture and sample barcoding. Together with our colleagues at the Dana-Farber Cancer Institute and Broad Institute, we have generated preliminary data showing the feasibility and promise of each of these components, thereby raising the possibility of comprehensive tumor sequencing at a low per-sample cost. Accordingly, in this R33 application we will implement a transformative platform for systematic tumor genomic profiling, which we call "MPS-OncoMap." To accomplish this we will optimize the methodology for sample barcoding technology, solution-phase exon capture, and single-molecule sequencing to enable robust mutation profiling (base mutations, amplifications, and deletions) across ~150 cancer genes in at least 12 tumor samples simultaneously. We will test the performance of MPS-OncoMap using DNA from cancer cell lines, frozen tumors, and formalin-fixed, paraffin-embedded tumor tissue for which the "ground truth" is known for multiple genetic alterations. Finally, we will implement MPS-OncoMap at production scale to enable systematic analyses for many translational oncology applications. Achieving these Aims may inform a definitive path to comprehensive tumor genomic profiling with far-reaching impact in the translational and clinical oncology arena. PUBLIC HEALTH RELEVANCE: High-Throughput Cancer Gene Mutation Profiling y Massively Parallel Sequencing PIs: Levi A. Garraway, M.D., Ph.D. and Laura E. MacConaill, Ph.D. Many cancers are caused by mutations in the DNA that give rise to altered proteins or cellular processes. The use of anticancer drugs targeting such proteins can lead to significant clinical benefit, but it remains impractical to profile the tumor of each cancer patient for the presence of such alterations. This application seeks to adapt a powerful new sequencing technology together with other innovations to make it possible to detect many informative cancer gene mutations in clinical tumor specimens. Once implemented, this approach could speed the advent of personalized cancer medicine.

DESCRIPTION (provided by applicant): Cancer remains the second leading cause of death in the United States. Furthermore, cancer is noteworthy because it is primarily a genomic disease: most tumors arise and persist because of a constellation of genomic changes that dysregulate cell growth and survival. Germline variants may also confer increased disease risk or thwart cancer treatment options by altering drug metabolism. The transformative medical potential of cancer genomic information has been made clear by the growing number of targeted agents that show remarkable efficacy against tumors whose salient genetic events confer heightened therapeutic vulnerability. Some mutations also identify tumors for which a therapy will be futile o even harmful. Many cancer genes harbor potentially actionable mutations at variable frequencies across a wide range of tumor types. These observations provide a compelling rationale for a paradigm wherein all therapeutically relevant tumor genomic alterations might be presented to physicians in a manner that guides personalized treatment. At first blush, translating the cancer genome for clinical use seems straightforward (Figure 1): (i) characterize the genome by massively parallel sequencing; (ii) filter this data through a compendium of available drugs/ targets; and (iii) present an annotated list to expert clinicians. However, multipe challenges must be addressed in order to bring this audacious goal to fruition. The first is a technical challenge: quality sequencing data must be obtained from limiting amounts of archival tumor tissue. The second is an analytical challenge: we must identify somatic and germline genomic changes with high accuracy, and distinguish driver events from the much larger set of passenger alterations. The third is a clinical challenge: we must achieve actionable data interpretation and develop a framework whereby genomic information promotes evidence based clinical trials and disease management choices. Finally, there is a psychosocial and ethical challenge: we must rigorously evaluate patients' and oncologists' experiences of clinical sequencing, and develop principled approaches to confront the myriad uncertainties that accompany the clinical genome era. We propose to establish a robust framework for generation, interpretation, and clinical implementation of cancer whole exome sequencing. To accomplish this, we have assembled world-class investigators from three of the world's top institutes for oncology and genomics: the Dana-Farber Cancer Institute (DFCI), the Broad Institute, and the Brigham and Women's Hospital (BWH). This team will leverage a major institutional partnership in personalized cancer medicine already in place at DFCI and BWH. After consent, patients will be enrolled into a clinical study (Project 1) wherein tumor and normal genomic DNA are procured and transferred to the Broad Institute for whole exome sequencing, analysis, and interpretation (Project 2). The resulting list of actionable alterations will be provided to the BWH diagnostic CLIA lab for validation and returned to the Project 1 clinical team to inform the care of cancer patients. The CLIA lab will independently query known actionable mutations using orthogonal approaches. In parallel, we will conduct longitudinal surveys and qualitative interviews of patients and their oncologists at various points surrounding the informed consent, data delivery and decision-making processes (Project 3). Upon completion, we will have configured a clinical formalism through which oncologists incorporate genomic information into their management plan and report the results to cancer patients and their families. The overall initiative will be jointly led by Drs. Levi Garraway and Pasi Janne. D. Garraway (Project 2 PI) is a cancer genome scientist and medical oncologist who has made pioneering advances at the interface of cancer genome characterization, targeted therapeutics, and personalized cancer medicine. Dr. Janne (Project 1 PI) is a translational oncologist who has made major discoveries highlighting the role of genomics in response and resistance to targeted anticancer agents. Dr. Steven Joffe (Project 3 PI) has made seminal contributions pertaining to the ethics of research, and Dr. Stacy Gray (Project 3 co-PI) is an outstanding'junior investigator focusing on communication/policy issues surrounding the return of genetic tests to cancer patients. The overall investigative team consists of world leaders in translational oncology, cancer genomics, clinical cancer genetics, computational biology, outcomes research, and research ethics. Together, these efforts will define a scalable model for the integration of clinical sequencing into cancer care.

DESCRIPTION (provided by applicant): In the United Sates, prostate cancer is the second leading cause of cancer-related deaths, and African- American men have long been known to exhibit a higher incidence and increased mortality compared to men of other ethnic groups. Although socioeconomic factors and unequal access to health care undoubtedly contribute to this disparity, differences in incidence and outcome persist after adjusting for these factors. Thus, tumor biological and/or genetic aspects may underpin adverse outcomes in African-American prostate cancer. To date, however, few studies have queried the molecular biology or somatic genetics of African-American prostate cancer in-depth, meaning that our understanding of the spectrum of factors influencing these disparities remains incomplete. The over-arching goal of this proposal is to undertake a definitive somatic genetic and functional characterization of African-American prostate cancer. To accomplish this goal, we will develop a targeted, massively parallel sequencing platform that interrogates the spectrum of recurrent alterations (base mutations, chromosomal copy number changes, and selected chromosomal rearrangements) relevant to prostate cancer. We will utilize this platform to perform tumor mutation profiling in a cohort of two hundred prostate tumor specimens obtained from African American men, leveraging knowledge from prostate cancer genome sequencing studies currently ongoing at the Broad Institute and elsewhere. Patterns of somatic alterations observed in this cohort will be compared to those of European-American prostate cancer. In addition, we will test top candidate genes affected by recurrently altered prostate cancer genomic loci for effects on the tumorigenic phenotype in primary prostate epithelial cells obtained from men of European or African descent. Once completed, this project should provide decisive insights into the spectrum of and functional relevance of somatic genetic alterations in African-American prostate cancer, thereby affirming or refuting the hypothesis that these biological factors undergird this pervasive cancer health disparity.

Although many men with advanced prostate cancer initially respond to androgen ablation therapy, the development of castration resistance is nearly inevitable. Progression to castration resistance leads to disease progression and, ultimately, death from prostate cancer. Despite intensive efforts, the genetic basis for castration resistance remains incompletely characterized. Similarly, the identification of genetic features that distinguish indolent from lethal disease has proved elusive. Advances in systematic genomic and functional technologies in recent years provide an unprecedented opportunity to make new scientific inroads into these longstanding challenges. The overarching goal of this research is to leverage these advances to address two cardinal questions in prostate cancer research: 1, the genetic determinants of indolent versus aggressive disease; and 2, a systematic understanding of mechanisms of resistance to castration-based therapy. These issues will be addressed through a series of complementary approaches. First, whole genome sequencing will be performed on >80 specimens obtained from castration-resistant bone metastases, or from primary tumors resected following neoadjuvant androgen blockade. Here, the goal is to identify recurrent genomic alterations that are preferentially associated with castration resistance. Next, we will develop a targeted massively parallel sequencing approach to enable high-throughput mutational profiling of mutations and rearrangements characteristic of clinically aggressive and/or castration resistant prostate cancer. This approach will be applied to 300 prostate tumor DNAs, thus providing a robust cohort in which to validate the observations from the whole genome sequencing efforts. The targeted approach may also provide a basis for new diagnostic modalities capable of stratifying prostate cancer patients for genomics-driven clinical trials. In addition, we will perform a genome-scale ORF screen to identify mechanisms of resistance to androgen ablation. Together, these studies should provide many new insights that propel new diagnostic and therapeutic options for men with clinically aggressive and castration-resistant prostate cancer.

Mutation of the BRAF oncogene is a common event in colorectal cancer (CRC). These mutations are associated with adverse outcome and insensitivity to epidermal growth factor receptor (EGFR) based therapy. Whereas RAF inhibitors have been successful in the treatment of BRAF-mutant malignant melanoma, response rates in BRAF-mutant CRC are surprisingly low. The basis for these disparate treatment outcomes remains incompletely understood. Preliminary studies from our groups found that suppression of the MAPK pathway by PLX4720 is incomplete in CRC lines. In some cases, this may be due to augmented EGFR-dependent signaling, but this does not explain all resistance in this setting. Our objective is to identify mechanisms operant in BRAF-mutant colorectal cancer that confer de novo resistance to RAF inhibitors, in hopes of enabling more efficacious therapeutics. First, differentially expressed genes linked to BRAF-mutant CRC will be identified by analysis of the TCGA dataset; these genes will be integrated with those that modify response to MAPK pathway inhibitors based on ongoing systematic functional screens. Top-ranking genes will be subjected to mechanistic studies to elucidate the molecular basis by which they confer resistance, in parallel, ongoing functional screens will be expanded to identify genes that are synthetic lethal with RAF inhibition in BRAF-mutant colorectal cells. Here, validation of leading candidates will be prioritized for genes that are potentially druggable-several candidates have already been nominated. Rational combinations of targeted agents with RAF inhibitors will be explored in cell culture and in xenograft models. Finally, clinical trials of combined RAF and MEK inhibitors will be performed at DF/HCC in an attempt to improve efficacy by enhancing suppression of the MAPK pathway and possibly prevent the emergence of drug-resistance. Tumor biopsies will be collected pre-treatment, on-treatment and post-progression, and whole exome and transcriptome sequencing will be used to identify genomic alterations that may drive resistance. Altogether, this work should provide a rigorous analysis of resistance to MAPK inhibitors and new therapeutic approaches to overcome them.

DESCRIPTION (provided by applicant): Advances in understanding the molecular basis of leukemia, together with the wealth of emerging innovative pharmaceutical compounds, have brought great opportunities to improve the clinical outcome of patients. To maximize the value of these discoveries, it is imperative to demonstrate our ability to molecularly characterizing individual tumors and apply this information to enroll patients onto therapeutic protocols that match distinct molecular profiles to specific targeting treatments-all within the context of an effective cancer therapeutic trials' system. To create an atmosphere for both research creativity and rapid clinical translation of novel discoveries into treatment approaches of adult leukemia patients, the Alliance for Clinical Trials in Oncology (hereafter called Alliance) Leukemia Correlative Science Committee (A-LCSC), the SWOG Leukemia Translational Medicine Subcommittee (S-LTMS) and the Broad Institute have elected to join forces and create an Integrated Translational Science Center for Leukemia (ITSC-L). The overall goal of the ITSC-L is to identify the genetic and molecular aberrations that characterize leukemic transformation, understand their contribution to therapeutic response or resistance, and utilize this information for the design of rational therapeutic trials that match specific molecular aberrations to emerging targeted therapies. These goals will be achieved through the carefully selection of Pilot Studies and Collaborative Projects that involve outstanding investigators working within Alliance and SWOG and those working outside the groups. Ultimately, these efforts should not only result in a significant improvement in the clinical outcome for leukemia patients, but are expected to generate cutting-edge findings applicable to other types of cancer. The Specific Aims of this Program are the following: Specific Aim 1. To create an Integrated Translational Science Center for Leukemia (ITSC-L) that will promote research among investigators working within the NCI National Clinical Trials Network (NCTN) system and outside the system by providing and coordinating funding, expertise, technology, tissue and data for innovative studies with clinical relevance; Specific Aim 2. To identify novel prognostic and predictive genetic and epigenetic markers and therapeutic targets for the development of innovative therapeutic paradigms that transform the current approach to leukemia patients and improve their cure rate; Specific Aim 3. To rapidly integrate high impact correlative science into early and late phase therapeutic trials o leukemia supported by the NCTN.

? DESCRIPTION (provided by applicant): Overcoming resistance to targeted therapy in cancer Project Summary The challenge of drug resistance represents a pervasive barrier that confounds the ultimate goal of cure or long-term control of metastatic cancer. Intensive studies of resistance to targeted therapies by our group in recent years have revealed three over-arching challenges. First, resistance is multifactorial: many individual resistance mechanisms thwart the efficacy of various targeted anticancer therapeutic regimens, and there is no evidence that their discovery has saturated. Second, resistance is heterogeneous: multiple different resistance mechanisms often arise in a given patient-even within a single tumor lesion1. Third, resistance is vastly under-sampled in the clinical arena: few paired pre-treatment and post-resistance biopsies are obtained clinically, and such tissues are only seldom subjected to systematic characterization2. The guiding hypothesis of this research is that the spectrum of resistance mechanisms for any given cancer therapeutic modality might coalesce onto a much smaller set of critical downstream effector nodes. Discerning the mechanisms operant within such points of coalescence should yield new insights into oncogenic dependencies and illuminate guiding principles for the design of novel therapeutic combinations. In recent years, we have evaluated this coalescence hypothesis by systematically characterizing mechanisms of resistance to MAP kinase pathway inhibition in BRAF-mutant melanoma and other oncogene- driven cancers. Indeed, many individual MAP kinase resistance mechanisms may coalesce at the level of transcriptional (or chromatin) regulation. This convergence re-engages core MAPK transcriptional program(s) or alternative (ERK-independent) transcriptional programs arising from bypass signaling or germane to pathway-indifferent cell states. Accordingly, one objective of this work is to define the convergent downstream outputs elaborated by MAP kinase inhibitor resistance mechanisms, and the factors that govern them. In parallel, we will characterize the coalescence of resistance mechanisms to targeted therapeutics in other cancers. Finally, we will describe drug-resistant and drug-indifferent cell states in treatment-refractory tumors. Detailed characterization of resistance categories and the mechanistic coalescence implied therein may reveal fundamental new insights into the nature of cancer dependencies and their evolution during tumor progression and treatment. Insights gleaned from this research may aid the design of higher-order therapeutic combinations that attack multiple tumor dependencies and resistance nodes. This framework for studying the mechanistic coalescence that underpins drug resistance is applicable to many cancer types. Therefore, these efforts could offer guiding principles that become generalizable across many tumor types and therapeutic modalities.

? DESCRIPTION (provided by applicant): The rapid evolution of cancer genome technology and computational analysis has engendered many fundamental cancer discoveries, thus transforming the scientific and clinical landscape in less than a decade. Increasingly, genetic alterations revealed by tumor genomic profiling guide diagnosis, treatment, and investigation in cancer. Despite these advances, many genomic technologies are unable to demonstrate drug targets or tumor-mediated drug resistance mechanisms that are not DNA-encoded. Furthermore, profiling approaches of bulk tumor samples only provide average signatures that do not reflect different tumor components and intrinsic heterogeneity of individual cell populations or cells. Emerging single-cell profiling technologies such as single-cell transcriptome analysis could overcome several challenges and provide a plethora of translational discovery opportunities. We recently provided proof- of-concept for application of single-cell RNA-seq in patient-derived tumor samples. To apply this technology more broadly in the translational oncology arena, we propose to create, optimize and implement a single-cell RNA-seq platform that can be deployed as a translational tool in the clinical oncology arena. In preliminary studies, we sequenced ~300 single cells from several melanoma tumors, including cancer and corresponding tumor-infiltrating cells (TILs). Transcriptome analysis revealed expression of key markers of melanoma, such as MITF and SOX10, and a stem-ness signature occurring in a fraction of melanoma cells. T cells expression reflected a spectrum of functional states, ranging from naïve to highly anergic (`exhausted') cells. These preliminary results demonstrate that we have created each necessary component for an end-to-end (clinic-to- bench) workflow, which yields meaningful single-cell data. The goal of this research is to optimize current protocols to create standard operating procedures (SOPs) and merge individual components into a standardized workflow. From our clinical colleagues, we will receive clinical specimens (tumors, biopsies, malignant effusions) from diverse tumor types, including melanoma, lung, breast and ovarian cancer. We will deploy experimental protocols to extract disaggregated individual cells from each sample type and state-of-the-art single cell RNA-Seq to profile each cell. Existing computational pipelines will be optimized to best serve high-throughput single-cell analysis, and once standardized will be made publicly available via an existing online platform. Upon completion of this project, we expect to have created a robust, multi-component workflow for single-cell transcriptome analysis applicable across cancer and sample types, and to render this technology accessible to the entire oncology community.

Although many men with advanced prostate cancer initially respond to androgen ablation therapy, the development of castration resistance is nearly inevitable. Progression to castration resistance leads to disease progression and, ultimately, death from prostate cancer. Despite intensive efforts, the genetic basis for castration resistance remains incompletely characterized. Similarly, the identification of genetic features that distinguish indolent from lethal disease has proved elusive. Advances in systematic genomic and functional technologies in recent years provide an unprecedented opportunity to make new scientific inroads into these longstanding challenges. The overarching goal of this research is to leverage these advances to address two cardinal questions in prostate cancer research: 1, the genetic determinants of indolent versus aggressive disease; and 2, a systematic understanding of mechanisms of resistance to castration-based therapy. These issues will be addressed through a series of complementary approaches. First, whole genome sequencing will be performed on >80 specimens obtained from castration-resistant bone metastases, or from primary tumors resected following neoadjuvant androgen blockade. Here, the goal is to identify recurrent genomic alterations that are preferentially associated with castration resistance. Next, we will develop a targeted massively parallel sequencing approach to enable high-throughput mutational profiling of mutations and rearrangements characteristic of clinically aggressive and/or castration resistant prostate cancer. This approach will be applied to 300 prostate tumor DNAs, thus providing a robust cohort in which to validate the observations from the whole genome sequencing efforts. The targeted approach may also provide a basis for new diagnostic modalities capable of stratifying prostate cancer patients for genomics-driven clinical trials. In addition, we will perform a genome-scale ORF screen to identify mechanisms of resistance to androgen ablation. Together, these studies should provide many new insights that propel new diagnostic and therapeutic options for men with clinically aggressive and castration-resistant prostate cancer.