Developmental and MAPK-responsive transcription factors regulate distinct malignant cell states and associated genetic dependencies in pancreatic cancer
TL;DR
The authors apply ARACNe and metaVIPER to published scRNA-seq datasets to characterize pancreatic cancer subtypes, identifying six distinct cell states as well as their mechanistic determinants.
Abstract
There is broad consensus that the malignant epithelial cells of human pancreatic ductal adenocarcinoma (PDA) comprise multiple, molecularly distinct states. Yet precise characterization of how these are regulated—including their mechanistic determinants, dependencies, plasticity and functional properties—remains elusive. Single-cell master regulator (MR) analysis of multiple PDA cohorts identified malignant cells in three co-existing, molecularly distinct developmental lineage states, with distinct histopathological morphologies and spatial architecture. These include a poorly differentiated lineage driven by epithelial–mesenchymal-transition-related MRs and two well-differentiated states driven by gastrointestinal epithelial development and pancreatic development MRs, respectively. Furthermore, each state comprises two epigenetically distinct substates with low versus high MAPK signaling activity. Barcode-based lineage tracing confirmed both spontaneous and treatment-dependent cross-state plasticity. Furthermore, loss-of-function studies confirmed state-specific MR essentiality, while their ectopic expression effectively reprogrammed cell state, in vitro and in vivo, thus providing a mechanism-based foundation for PDA heterogeneity and a roadmap for pharmacological targeting. The authors apply ARACNe and metaVIPER to published scRNA-seq datasets to characterize pancreatic cancer subtypes, identifying six distinct cell states as well as their mechanistic determinants. Functional experiments in vivo and in vitro demonstrate how these inferred states display plasticity and associate with functional dependencies.