Abstract Pancreatic cancer is a challenging disease which lacks a robust precision oncology framework. Prior work has demonstrated that transcriptional subtypes in pancreatic cancer can stratify patients by response to chemotherapy and overall survival. These include classical and basal-like as the predominant subtypes in untreated disease, as well as a transdifferentiated neural-like malignant subtype enriched after standard-of-care chemotherapy. Nonetheless, our understanding of transcriptional subtypes remains critically limited by genetic and culture-based confounders in preclinical models. Furthermore, most murine pancreatic cancer models do not fully reflect the classical vs. basal-like divide observed in humans, preventing faithful in vivo investigation. Collectively, these challenges have hindered the study of and development of therapeutic strategies against these transcriptional subtypes. To enable systematic study of the classical, basal-like and neural-like subtypes, we developed isogenic KP (KrasG12D/+;Trp53FL/FL) organoids and leveraged CRISPR activation to endogenously overexpress state-specific transcription factors. Our data revealed that Gata6 and dNTrp63 drove classical and basal-like identity, respectively, supporting prior studies. Furthermore, through in silico approaches, we identified candidate drivers of the neural-like state, including Glis3, a zinc finger protein that mediates neuroendocrine fate during pancreatic development. We confirmed that Glis3 overexpression results in neural-like transdifferentiation in cancerous organoids through RNA-seq and proteomics, and that GLIS3 knockout abrogates neural-like identity in human cell lines. To study their malignant potential in vivo, we performed orthotopic transplants into murine pancreata, and observed that Glis3 and dNTrp63 accelerated tumor formation and progression, corresponding to clinical outcomes observed in patients. These models may therefore provide a means for in vivo interrogation of subtype biology and vulnerabilities. Finally, to model and predict therapeutic sensitivities, we performed ex vivo treatment assays with chemotherapy and emerging KRASG12D inhibitors. A selection assay showed that basal-like and neural-like organoids were starkly enriched by multi-agent chemotherapy, while classical organoids were depleted. Finally, we learned that the neural-like subtype exhibits differential sensitivity to KRASG12D inhibition through an RTK-associated mechanism. Taken together, we provide a foundational system for interrogating subtype-specific vulnerabilities in pancreatic cancer. Citation Format: Jimmy A. Guo, Jennifer Su, Carina Shiau, Ananya Jambhale, Annan Yang, Westley Wu, Junning Wang, Connor Hennessey, Patrick Yu, Brendan Parent, Giselle Uribe, Julien Dilly, Laleh Abbassi, Qijia Yu, Arnav Mehta, David Ting, Brian Wolpin, William Freed-Pastor, Joseph Mancias, Tyler Jacks, William L. Hwang, Andrew J. Aguirre. GLIS3 drives a neural-like malignant state enriched after neoadjuvant treatment in pancreatic cancer. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5775.
Read full abstract