Discrete bladder cancer molecular subtypes exhibit differential clinical aggressiveness and therapeutic response, which may have significant implications for identifying novel treatments for this common malignancy. However, research is hindered by the lack of suitable models to study each subtype. To address this limitation, we classified bladder cancer cell lines into molecular subtypes using publically available data in the Cancer Cell Line Encyclopedia (CCLE), guided by genomic characterization of bladder cancer by The Cancer Genome Atlas (TCGA). This identified a panel of bladder cancer cell lines which exhibit genetic alterations and gene expression patterns consistent with luminal and basal molecular subtypes of human disease. A subset of bladder cancer cell lines exhibit in vivo histomorphologic patterns consistent with luminal and basal subtypes, including papillary architecture and squamous differentiation. Using the molecular subtype assignments, and our own RNA-seq analysis, we found overexpression of GATA3 and FOXA1 cooperate with PPARɣ activation to drive transdifferentiation of a basal bladder cancer cells to a luminial phenotype. In summary, our analysis identified a set of human cell lines suitable for the study of molecular subtypes in bladder cancer, and furthermore indicates a cooperative regulatory network consisting of GATA3, FOXA1, and PPARɣ drive luminal cell fate.

Muscle-invasive bladder cancers are characterized by their distinct expression of luminal and basal genes, which could be used to predict key clinical features such as disease progression and overall survival. Transcriptionally, FOXA1, GATA3, and PPARG are shown to be essential for luminal subtype-specific gene regulation and subtype switching, while TP63, STAT3, and TFAP2 family members are critical for regulation of basal subtype-specific genes. Despite these advances, the underlying epigenetic mechanisms and 3D chromatin architecture responsible for subtype-specific regulation in bladder cancer remain unknown. Result We determine the genome-wide transcriptome, enhancer landscape, and transcription factor binding profiles of FOXA1 and GATA3 in luminal and basal subtypes of bladder cancer. Furthermore, we report the first-ever mapping of genome-wide chromatin interactions by Hi-C in both bladder cancer cell lines and primary patient tumors. We show that subtype-specific transcription is accompanied by specific open chromatin and epigenomic marks, at least partially driven by distinct transcription factor binding at distal enhancers of luminal and basal bladder cancers. Finally, we identify a novel clinically relevant transcription factor, Neuronal PAS Domain Protein 2 (NPAS2), in luminal bladder cancers that regulates other subtype-specific genes and influences cancer cell proliferation and migration. Conclusion In summary, our work identifies unique epigenomic signatures and 3D genome structures in luminal and basal urinary bladder cancers and suggests a novel link between the circadian transcription factor NPAS2 and a clinical bladder cancer subtype.

Muscle-invasive bladder cancers have recently been characterized by their distinct expression of luminal and basal genes, which could be used to predict key clinical features such as disease progression and overall survival. For example, FOXA1, GATA3, and PPARG have been shown to be essential for luminal subtype-specific regulation and subtype switching, while TP63 and STAT3 are critical for basal subtype bladder cancer. Despite these advances, the underlying epigenetic mechanism and 3D chromatin architecture for subtypespecific regulation in bladder cancers remains largely unknown. Here, we determined the genome-wide transcriptome, enhancer landscape, TF binding profiles (FOXA1 and GATA3) in luminal and basal subtypes of bladder cancers. Furthermore, we mapped genome-wide chromatin interactions by Hi-C in both bladder cancer cell lines and primary patient tumors, for the first time in bladder cancer. We showed that subtype-specific transcription is accompanied by specific open chromatin and epigenomic marks, at least partially driven by distinct TF binding at distal-enhancers of luminal and basal bladder cancers. Finally, we identified a novel clinically relevant transcriptional factor, Neuronal PAS Domain Protein 2 (NPAS2), in luminal bladder cancers that regulates other luminal-specific genes (such as FOXA1, GATA3, and PPARG) and affects cancer cell proliferation and migration. In summary, our work shows a subtype-specific epigenomic and 3D genome structure in urinary bladder cancers and suggested a novel link between the circadian TF NPAS2 and a clinical bladder cancer subtype.