Abstract 210: Uncovering a novel layer of complexity in the architecture of pancreatic cancer

Abstract

Abstract Single cell technologies have described heterogeneity across tissues, but the spatial distribution and forces that drive single cell phenotypes has not been well defined. We integrated single cell RNA-seq (scRNA-Seq) and phospho mass cytometry (CyTOF) with high-content digital imaging to investigate tumor heterogeneity of human pancreatic cancer (PDAC) using a novel tumor-architecture approach. Patient-derived PDAC cells co-cultured at different ratios with cancer associated fibroblasts (CAFs) were found to heterogeneously acquire proliferative (PRO) and invasive (EMT) phenotypes. scRNA-Seq enabled the identification of a novel cell phenotype with upregulation of both PRO and EMT programs, which we named the double positive (DP) phenotype. Functional studies confirmed the heterogeneous activation of PRO and EMT programs with different patterns of tumor growth and metastasis in mice orthotopically xenografted with distinct PDAC:CAF ratios. In a time-course experiment of PDAC cells with CAF conditioned media (CM), mass spectrometry-based phosphoproteomics demonstrated selective enrichment of MAPK and STAT3 pathways. This was validated with mass cytometry (CyTOF) at the single cell level in our cell line model as well as from primary PDAC tumors, which identified the dual enrichment of MAPK and STAT3 signaling in the DP cell population. Functional validation of MAPK and STAT3 signaling was performed with small molecule inhibitors of MAPK (Trametinib) and STAT3 (pyrimethamine) in our cell line models. To evaluate the generalizability of these cell subpopulations, we performed RNA-ISH for EMT (FN1) and PRO (MKI67) across 195 human PDAC primary tumors and scored 319,626 cancer cells revealing significant heterogeneity of PRO, EMT, and DP cells among patients. Single-cell analysis within the context of tissue architecture revealed these cells were grouped together in discrete tumor glands. Using each of the 3 cell types (PRO, EMT, and DP), we classified 8 distinct types of tumor glands in these primary tumors. This provided the ability to define inter and intra-tumor heterogeneity at the tumor gland level, which provides an extra layer of tumor cell organization not appreciable with single cell analysis alone. We showed that tumor glands are independent functional units in human PDACs carrying distinct prognostic information. In fact, by comparing their prognostic power we noticed that some tumor glands are positively linked with worsened survival (Type I p=0.003, Type II p=0.04, Type II p=0.001 and Type VII p=0.02), while Type III glands are associated with a favorable patient prognosis (p=0.01, log-rank test). In conclusion, our study showed for the first time at this scale level that integrating single-cell technologies with tissue architectural information provides a more comprehensive landscape of intra-tumoral cancer cell heterogeneity that has implications on PDAC cell behavior and patient outcomes. Citation Format: Matteo Ligorio, Srinjoy Sil, Jose Malagon-Lopez, Sandra Misale, Murat Karabacak, Linda Nieman, Shyamala Maheswaran, Daniel A. Haber, Andrew L. Warshaw, Carlos Fernandez-Del Castillo, Cristina R. Ferrone, Wilhelm Haas, Martin Aryee, David T. Ting. Uncovering a novel layer of complexity in the architecture of pancreatic cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 210.