Cell states and therapeutic targets in pancreatic cancer

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is an aggressive cancer wherein treatment resistance is common and survival is poor. Prior transcriptional profiling studies have revealed two major cell states in untreated human PDAC – classical and basal-like – with differential prognoses and clinical responses to treatment. As such, transcriptional cell states have been the subject of considerable investigation in the pancreatic cancer field over the past decade. While the basal-like vs. classical classification can inform prognosis, many patients with localized disease receive upfront chemotherapy prior to surgery, and most patients enrolled on clinical trials have likely received prior chemotherapy. We therefore sought to understand whether and how treatment alters the transcriptional cell state taxonomy and performed single-nucleus RNA-seq and spatial mRNA profiling on primary tumors treated with or without neoadjuvant chemotherapy and radiation. In doing so, we discovered the existence of a novel ‘neural-like progenitor’ cancer cell state which was significantly enriched in treated cancer cells, and found that it is regulated by GLIS3, a transcription factor canonically involved in neuroendocrine lineage specification. We next sought to functionally investigate the biology of the classical, basal-like, and neural-like cell states through the development of ex vivo and in vivo isogenic models driven by transcription factor overexpression. This revealed distinct morphologies, in vivo growth kinetics, metastatic proclivities, and drug sensitivities associated with each cell state. Furthermore, cell-intrinsic gene expression patterns were associated with cell-extrinsic microenvironmental features; for instance, the neural-like cancer state was enriched with a gene signature for perineural invasion. Finally, we leveraged functional genetic screens in cell lines and patient-derived organoids to survey novel or underexplored therapeutic targets for this devastating disease. Together, our findings improve our understanding of the biology of cancer cell states and provide guidance for future target validation efforts in pancreatic cancer.