Natalie Geyer

Despite recent advances in oncology, the prognosis of patients with pancreatic cancer remains dismal. The main reason for the poor clinical outcome is the high propensity of pancreatic cancer to spread to distant organs, especially to the liver. However, little is known about the biology of pancreatic cancer liver metastases (PCLM). It has been established in primary pancreatic tumors that interactions with tumor cells and their microenvironment have strong impact on tumor cell aggressiveness. However, despite the clinical importance of metastases, little is known about the role of these interactions in PCLM. In this project, I aim to shed light on tumor-microenvironment interactions in PCLM by charting metastatic invasion on the single cell level. Specifically, I will decipher the key cellular interactions and underlying signaling pathways governing invasion in PCLM by single-cell RNA sequencing (sc-RNA seq), combined with novel computational methods to deconvolute cell-cell interactions at the metastatic invasion front. In preliminary experiments, I have used sc-RNA seq of murine PCLM to create a single cell atlas covering all major cell populations, including tumor cells and the diverse non-neoplastic cell types of the liver. In order to identify physically interacting cells in intact tissue, I went on to use a novel computational method, Cellular Interaction by Multiplet signaling (CIM-seq), that uses machine learning to disentangle cell multiplets. My preliminary data show that CIM-seq can identify significant interactions between tumor cell classes and cells of the tumor microenvironment

most importantly, my results provide proof-of-principle for the potential of CIM-seq to disentangle cellular interactions in a tissue as complex as liver metastases. In part A of this project, I propose to complete our sc-RNA seq dataset, and to use CIM-seq to specifically chart interactions at the invasion front of murine PCLM. Next, I will use multiplex RNA in situ hybridization and multiplex immunohistochemistry on human samples of PCLM to spatially validate my results in patient material. In part B, I will follow a functional approach and use genetically modified mice to assess key signaling pathways underlying the interactions between stromal and tumor cells. I will begin with modifying the activity of Hedgehog signaling, which regulates paracrine tumor-microenvironment crosstalk in primary pancreatic cancer. Specifically, I will activate Hedgehog signaling in the stroma of murine PCLM to investigate its impact on tumor growth and invasion. In the long run, I aim to develop a CRISPR/Cas9-based mouse model to modify signaling pathways identified by our sc-RNA seq data that may influence tumor aggressiveness in a highly versatile manner. The results of these project are expected to yield important new insights on metastatic invasion of PCLM, and they could ultimately inform the development of novel treatment strategies for this devastating disease.