Abstract Untreated prostate cancers rely on androgen receptor (AR) signaling for growth and survival, forming the basis for the initial efficacy of androgen deprivation therapy (ADT). Yet the disease can relapse and progress to a lethal stage termed castration-resistant prostate cancer (CRPC). Reactivation of AR signaling represents the most common driver of CRPC growth, and next-generation AR signaling inhibitors (ARSIs) are now used in combination with ADT as first-line therapy. However, ARSIs can result in selective pressure, thereby generating AR-independent tumors. The transition from AR dependence frequently accompanies a change in a phenotype resembling developmental transdifferentiation or “lineage plasticity”. Neuroendocrine prostate cancer, which lacks a defined pathologic classification, is the most studied type of lineage plasticity. However, most AR-null tumors do not exhibit neuroendocrine features and are classified as “double-negative prostate cancer”, the drivers of which are poorly defined. Lineage plasticity studies in CRPC are limited by the lack of genetically defined patient-derived models that recapitulate the disease spectrum. To address this, we developed a biobank of organoids generated from patient biopsies to study the landscape of metastatic CRPC and allow for functional validation assays. Proteins called transcription factors (TFs) are drivers of tumor lineage plasticity. To identify the key TFs that drive the growth of AR-independent tumors, we integrated epigenetic and transcriptomic data generated from CRPC models. We generated ATAC-seq (assay for transposase-accessible chromatin sequencing) and RNA-seq data from 22 metastatic human prostate cancer organoids, six patient-derived xenografts (PDXs), and 12 derived or traditional cell lines. We classified the 40 models into four subtypes and predicted key TFs of each subtype. Besides the well-characterized AR-dependent (CRPC-AR) and neuroendocrine subtypes (CRPC-NE), we identified two novel AR-negative/low groups, including a Wnt-dependent subtype (CRPC-WNT), driven by TCF/LEF TFs, and a stem cell-like (SCL) subtype (CRPC-SCL), driven by the AP-1 family of TFs. To apply the subtype classification to patient samples, we derived RNA-seq signatures from the organoids and applied them to 366 patient samples from two independent CRPC cohorts. The generated signatures recapitulated the four-subtype classification and revealed that CRPC-SCL is the second most prevalent group. Patients from CRPC-SCL are also associated with shorter time under ARSI treatment compared to CRPC-AR, indicating that the ARSI treatments were less effective for CRPC-SCL patients. Additional chromatin immunoprecipitation sequencing (ChIP-seq) analysis indicated that AP-1 (FOSL1) collaboratively binds with TEAD and transcription coactivators, YAP and TAZ. Knocking down of AP-1 (FOSL1), YAP/TAZ decreased cell growth of CRPC-SCL and showed a decrease of chromatin accessibility at CRPC-SCL-specific open chromatin sites and down-regulation of YAP/TAZ target gene expression. In addition, the expression of AP-1 (FOSL1) decreased upon YAP/TAZ knockdown suggesting a positive feedback loop as well as YAP/TAZ as actional targets in CRPC-SCL. We used two small-molecule inhibitors, verteporfin and T-5224, that act on the YAP/TAZ/AP-1 pathway for their potential use as therapeutics for CRPC-SCL tumors, both inhibited the growth of samples from CRPC-SCL but not CRPC-AR. By overexpressing an AP-1 family gene (FOSL1) in AR-high cells, we observed an increase in chromatin accessibility at CRPC-SCL-specific open chromatin sites as well as significant up-regulation of CRPC-SCL signature genes, suggesting that AP-1 functions as a pioneering factor and master regulator for CRPC-SCL. All this work was recently published in Science (Tang, Xu et al. Science, 2022) where I am the co-first author. In summary, by using a diverse biobank of organoids, PDXs, and cell lines that recapitulate the heterogeneity of metastatic prostate cancer, we created a map of the chromatin accessibility and transcriptomic landscape of CRPC. We validated the CRPC-AR and CRPC-NE subtypes and report two novel subtypes of AR-negative/low samples, CRPC-SCL and CRPC-WNT, as well as their respective key TFs. Additional analysis revealed a model in which YAP, TAZ, TEAD, and AP-1 function together and drive oncogenic growth in CRPC-SCL samples. In addition, we proposed small inhibitors of YAP and TAZ that can potentially be used to treat CRPC-SCL patients. Overall, our results show how the stratification of CRPC patients into four subtypes using their transcriptomes can potentially inform appropriate clinical decisions. Citation Format: Fanying Tang, Duo Xu, Shangqian Wang, Chen Khuan Wong, Alexander Martinez-Fundichely, Cindy J. Lee, Sandra Cohen, Jane Park, Corinne E. Hill, Kenneth Eng, Rohan Bareja, Teng Han, Eric Minwei Liu, Ann Palladino, Wei Di, Dong Gao, Wassim Abida, Shaham Beg, Loredana Puca, Maximiliano Meneses, Elisa de Stanchina, Michael F. Berger, Anuradha Gopalan, Lukas E. Dow, Juan Miguel Mosquera, Himisha Beltran, Cora N. Sternberg, Ping Chi, Howard I. Scher, Andrea Sboner, Yu Chen, Ekta Khurana. Chromatin profiles classify castration-resistant prostate cancers suggesting therapeutic targets. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr NG10.