Single-cell data for FGFR3-driven gene regulatory network analysis reveals a pro-tumoral role for p63 in luminal bladder tumors

FGFR3 (Fibroblast Growth Factor Receptor 3) is one of the most frequently altered genes in bladder cancer, primarily through activating mutations that drive oncogenesis and are enriched in luminal tumors. However, the underlying gene regulatory network (GRN) remains poorly characterized. Here, we inferred an FGFR3-mutated GRN using a bottom-up bioinformatics approach, integrating transcriptomic data from bladder cancer cell lines, FGFR3-mutated tumors, and FGFR3-perturbation experiments in human and mouse models. Using CRISPR-Cas9 screen public data, we identified transcription factors (TFs) from this GRN that regulate the viability of FGFR3-mutated cells, with a focus on p63 (TP63). We showed that FGFR3 activation upregulates p63 in patient-derived xenografts and cell lines, while single-cell RNA sequencing revealed heterogeneous p63 activation associated with basal differentiation. Functional studies, including TP63 knock-down in FGFR3-dependent in vitro and in vivo models and RNA-seq along with p63 ChIP-seq, demonstrated that p63 directly promotes cell proliferation and migration, and uncovered a positive feedback loop between FGFR3 and p63. Together, these findings identified p63 as a pro-tumorigenic regulator in FGFR3 mutated tumors despite their luminal differentiation and provide a detailed FGFR3-driven GRN, offering insights into FGFR3-induced oncogenic dependency and potential strategies to circumvent resistance to FGFR inhibitors. Overall design: MGH-U3 and RT112 cells were treated with DMSO or 100 nM erdafitinib for 38 h and patient-derived tumor xenografts (PDXs) from the F659 model bearing FGFR3 mutations ?were sequenced via single-cell RNA-seq. PDXs from F659 models, harboring an FGFR3-S249C mutation, were obtained as detailed in the publication attached (PMID: 29463565), they were generated by engrafting tumor tissues directly obtained from patients. Fresh tumors were used for single-cell RNA-seq analysis. The sequencing output was processed using Cell Ranger from 10x Genomics?.