Abstract A007: Pancreatic cancer comprises co-existing transcriptional states regulated by distinct master regulator programs

Abstract

Despite extensive efforts, reproducible assessment of pancreatic ductal adenocarcinoma (PDA) heterogeneity and plasticity at the single cell level remains elusive. Systematic, network-based analysis of single cell RNA-seq profiles showed that most PDA tumors comprise three coexisting lineages whose aberrant transcriptional state is mechanistically controlled by distinct regulatory programs. These lineages were characterized by the aberrant activation of either gastrointestinal lineage markers (GLS), transcriptional effectors of morphogen pathways (MOS) and acinar to ductal metaplasia markers (ALS). Each lineage was characterized by cells in two different cell states determined by the differential activation of MEK signaling (M+/M-) and high cellular plasticity. These states were confirmed in multiple cohorts, cell lines, PDX models and harmonized with bulk profile analyses. Master regulators (MRs) of GLS and MOS state were predictive of patient’s survival in bulk profiles. Cross-state plasticity was confirmed by lineage tracing assays, while pooled CRISPR/Cas9 assays confirmed the essentiality of identified MR proteins. Finally, mechanistic MR-mediated cell state control was confirmed by MR expression-mediated reprogramming of MOS cells to a GLS state. Our work provided a mechanistic model of pancreatic cancer heterogeneity and testable hypothesis to target cell state-specific pancreatic cancer dependencies. Citation Format: Pasquale Laise, Mikko Turunen, Hans Carlo Maurer, Alvaro Curiel Garcia, Ela Elyada, Bernhard Schmierer, Lorenzo Tomassoni, Jeremy Worley, Mariano J. Alvarez, Jordan Kesner, Xiangtian Tan, Somnath Tagore, Ester Calvo Fernandez, Kelly Wong, Alexander L. E. Wang, Sabrina Ge, Alina C. Iuga, Aaron T. Griffin, Winston Wong, Gulam A. Manji, Faiyaz Notta, David A. Tuveson, Kenneth P. P. Olive, Andrea Califano. Pancreatic cancer comprises co-existing transcriptional states regulated by distinct master regulator programs [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer; 2022 Sep 13-16; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2022;82(22 Suppl):Abstract nr A007.

Morphology-guided transcriptomic analysis of human pancreatic cancer organoids reveals microenvironmental signals that enhance invasion

“…Using the DECODER pipeline (33) to infer associations with widely recognized classical vs. basal transcriptomic subtyping scheme for primary PDAC tissue originally published by Moffitt et al (8), two of three tG2 invasive organoid samples were assigned as the classical subtype, whereas the other invasive organoids were classified as the basal subtype (Figure 3B). Similarly, a recently proposed PDAC subtyping scheme using sets of transcriptional regulators (34) classified the tG1 as in the morphogenic state and both tG2 and tG3 as in the lineage state. Lastly, we applied gene signatures previously derived from PDAC organoids distinguished by their speed of metastatic progression in murine intraductal xenografts (35).…”

Identification and Pharmacological Targeting of Treatment-Resistant, Stem-like Breast Cancer Cells for Combination Therapy

“…A key question in network-based analyses is whether-similar to what has been shown in other contexts 75,76,96 -candidate MRs may comprise hyper-connected, autoregulated modules providing coordinated, homeostatic cell state regulation. For this purpose, we assessed whether metaVIPER-inferred CSLC MRs were statistically significantly enriched in protein-protein and transcriptional interactions-as reported in PrePPI 97 , STRING 98 , and ARACNe-based networks-compared to an equivalent number of same-class proteins selected at random.…”

“…The high validation rate of VIPER-inferred MRs in the CROP-seq analysis suggests that MR-inverter drugs capable of inhibiting and activating the most positive and negative MRs, respectively, should induce CSLC differentiation, thus increasing their sensitivity to chemotherapy. Indeed, MR-mediated reprogramming of cell state has already been validated in multiple contexts, from dedifferentiation 84 , to reprogramming 96,104 and trans-differentiation 85,105 . For this purpose, we leveraged the OncoTreat algorithm, which has proven highly effective in discovering MR-inverter drugs that were extensively validated in vivo, based on MR proteins inferred by VIPER analysis of both bulk [17][18][19] and single-cell profiles 20,21 .…”