Abstract NG06: CHD1-loss confers AR targeted therapy resistance via promoting cancer heterogeneity and lineage plasticity

Abstract BACKGROUND: Pharmacological targeting of driver alterations in cancer has resulted in many clinical successes but is limited by concurrent or novel genomic alterations. One potential explanation for this heterogeneity is the presence of additional genomic alterations which modify the degree of dependence on the targeted driver mutation. Metastatic prostate cancer (mPCa) serves as a relevant example, where the molecular target is the androgen receptor (AR) which functions as a lineage survival factor of luminal prostate epithelial cells. Next generation AR targeted therapies such as abiraterone, enzalutamide and apalutamide have significantly improved the survival of men with mPCa and achieved exciting clinical success. However, resistance to these therapies with disease progression is unfortunately inevitable, with intrinsic resistance noted in around 30% patients and acquired resistance in most patients. Therefore, there is an unmet need to understand the mechanism of therapy resistance to AR targeted therapies and identify novel therapeutic approach to prevent or reverse resistance. Previously, we have revealed that the deactivation of two genes, TP53 and RB1, confers AR targeted therapy resistance through a novel mechanism by which tumor cells acquire lineage plasticity and transit to a multi-lineage, progenitor-like state no longer dependent on AR. This lineage plasticity and resistance is enabled by the activation of SOX2 and is completely reversible by knocking down SOX2. This observation not only adds clarity to the mechanism of resistance, but also suggests that appropriate clinical interventions of lineage plasticity may be a potential avenue to overcome resistance. However, there is only 10% mPCa patients carrying homozygous deletions in both TP53 and RB1 loci, thus additional genomic alterations may be responsible for the resistance in other patients. METHODS: To gain functional insight into the genes impacted by the copy number alterations in mPCa, we screened 4234 short hairpin RNAs (shRNAs) targeting 730 genes often deleted in human prostate cancer (annotated from a survey of six prostate cancer genomic datasets) for hairpins that confer in vivo resistance to the antiandrogen enzalutamide, in a well credentialed enzalutamide-sensitive xenograft model LNCaP/AR. More than 350 resistant tumors emerged by 16 weeks of xenografting and the genomic DNA of these tumors were extracted and sequenced to determine the enrichment of specific shRNAs compared to the starting material. A classic probabilistic model RIGER-E was used to determine the significance of enrichment of each hairpins/genes. RESULTS: The chromodomain helicase DNA-binding protein 1 (CHD1) emerged as a top candidate, a finding supported by patient data showing that CHD1 expression is inversely correlated with clinical benefit from AR targeted therapy enzalutamide. CRISPR based depletion of CHD1 confers significant resistance to enzalutamide both in vitro and in vivo, supported by similar results from multiple human prostate cancer cell line models. To our surprise, we observed sustained inhibition of the canonical AR target genes, indicating that CHD1 loss might activate transcriptional programs that relieve prostate tumor cells from their dependence on AR by reprogramming away from their luminal lineage, as we have observed in the setting of combined loss of RB1 and TP53. Indeed, CHD1 loss led to global changes in open and closed chromatin, indicative of an altered chromatin state, with associated changes in gene expression. Integrative analysis of ATAC-seq and RNA-seq changes identified 22 transcription factors as candidate drivers of enzalutamide resistance. CRISPR deletion of four of these (NR3C1, BRN2, NR2F1, TBX2) restored in vitro enzalutamide sensitivity in CHD1 deleted cells. Independently derived, enzalutamide-resistant, CHD1-deleted subclones expressed elevated levels of 1 or more of these 4 transcription factors. This pattern suggests a state of chromatin plasticity and enhanced heterogeneity, initiated by CHD1 loss, which enables upregulation of distinct sets of genes in response to selective pressure. This concept is further supported by RNA-seq data from a mCRPC patients cohort, in which we examined the co-association of CHD1 levels with each of these four TFs across 212 tumors. Unsupervised clustering analysis of just these five genes identified five distinct clusters, four of which display relatively higher expression of either CHD1 or one or two of these four resistance TFs. Interestingly, we observed altered expression of many canonical lineage specific genes in the same panel of CHD1-deleted, enzalutamide resistant xenografts that displayed heterogenous upregulation of the four TFs, including consistent downregulation of luminal genes and upregulation of genes specify epithelial to mesenchymal transition (EMT). Furthermore, these upregulation of 4 resistance TFs, as well as the observed lineage switchs, are both rapid and reversible, suggesting a status of plasticity. Collectively, these results indicate that CHD1 loss establishes an altered chromatin landscape which, in the face of stresses such as antiandrogen therapy, enables resistant subclones to emerge through activation of alternative, non-luminal lineage programs that reduce dependence on AR. CONCLUSIONS: We demonstrated that loss of the chromodomain gene CHD1, a commonly deleted prostate cancer gene (in 15~20% patients), through global effects on chromatin, establishes a state of plasticity that accelerates the development of AR targeted therapy resistance through heterogeneous activation of downstream effectors, which mediated the transition away from luminal lineage identity and AR dependency. This model provides the first demonstration that early genomic lesions of critical epigenetic modulator promotes prostate cancer heterogeneity and lineage plasticity, consequently leading to the resistance to AR targeted therapy. Therefore, appropriate clinical intervention of these heterogenous resistance driver TFs, as well as the chromatin dysregulation, may be potential therapeutic avenues to prevent or reverse AR targeted resistance. Citation Format: Zeda Zhang, Chuanli Zhou, Xiaoling Li, Spencer Barnes, Su Deng, Elizabeth Hoover, Chi-Chao Chen, Young Sun Lee, Choushi Wang, Carla Tirado, Lauren Metang, Nickolas Johnson, John Wongvipat, Kristina Navrazhina, Zhen Cao, Wassim Abida, Amaia Lujambio, Sheng Li, Vankat Malladi, Charles Sawyers, Ping Mu. CHD1-loss confers AR targeted therapy resistance via promoting cancer heterogeneity and lineage plasticity [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr NG06.