Lineage plasticity is a major mechanism driving prostate cancer progression and antiandrogen therapy resistance. Deletions or mutations in phosphatase and tensin homolog (PTEN) and TP53 tumor suppressor genes have been linked to lineage plasticity in prostate cancer. Fusion-driven overexpression of the E-twenty-six transformation specific (ETS)-related gene (ERG), encoding an oncogenic transcription factor, is observed in approximately 50% of all prostate cancers, yet its role in prostate cell lineage determination remains elusive. Here we demonstrate that transgenic expression of prostate cancer-associated ERG blocks Pten and Trp53 mutation-induced decreased expression of Ar and its downstream target genes and loss of luminal epithelial cell identity in the mouse prostate. Integrative analyses of ERG chromatin-immunoprecipitation sequencing (ChIP-seq) and transcriptome data show that ERG suppresses expression of a subset of cell cycle-promoting genes and RB phosphorylation, which in turn causes repression of E2F1-mediated expression of non-epithelial lineage genes. Xenograft studies show that PTEN/TP53 double mutated prostate tumors are responsive to the cyclin-dependent kinase 4 or 6 (CDK4/6) inhibitor palbociclib, but resistant to the AR inhibitor enzalutamide, while ERG/PTEN/TP53 triple-mutated prostate tumors behave completely opposite. Our studies identify ERG and the repressed cell cycle gene signature as intrinsic inhibitors of PTEN/TP53 double mutation-elicited lineage plasticity in prostate cancer. Our findings also suggest that ERG fusion can be utilized as a biomarker to guide the treatment of PTEN/TP53-mutated, RB1-intact prostate cancer with either antiandrogen or anti-CDK4/6 therapies. Overall design: Prostate tissue from mice with 1) prostate specific PTEN deletion, p53 R172H mutation with loss of heterozygosity, or 2) prostate specific PTEN deletion, p53 R172H mutation with loss of heterozygosity and transgenic ERG expression were harvested at 4-5 months. RNA was isolated from tissue and RNA-seq experiments were then performed for both genotype samples in triplicates. Differentially expressed genes were identified by comparing genotype #1 and genotype #2.