Investigation of tumour/stroma crosstalk in pancreatic cancer by an advanced 3D co-culture model

Abstract: Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal malignancies all over the world. This dire situation is mainly attributable to the dense desmoplasia in the tumour microenvironment (TME) and almost complete resistance to conventional chemotherapy. Cancer-associated fibroblasts (CAFs) are the key component in the TME of PDAC, which organize the excessive desmoplasia and largely affect tumour biology and treatment response. Recently, the significant heterogeneity of CAFs, regarding their origins, spatial distribution and biological function, has been discovered. However, the crosstalk between tumour cells and CAFs has been less investigated, mostly due to the lack of appropriate study models. In this thesis, we aimed to develop advanced three-dimensional (3D) co-culture models of pancreatic tumour cells and pancreatic stellate cells (PSCs, the major source of CAFs in PDAC) to dissect the interactions between tumour and stroma in PDAC. In paper I, we first set up a heterospheroids model by directly co-culturing human PSCs (hPSCs) with human pancreatic tumour cells to study tumour/stroma crosstalk. Characterisation of the 3D co-culture model by immunohistochemistry and qRT-PCR, we discovered that the tumour cells had increased proliferation and an epithelial–mesenchymal transition phenotype upon co-culture with hPSCs. In addition, human PSCs also got activated towards a more myofibroblastic phenotype. Furthermore, a mixed-species heterospheroids model of PDAC cells and PSCs that allowed to detect the gene expression in specific cell types was developed, which confirmed the discoveries that we detected from the same-species heterospheroids model. In paper II, we analysed the tumour/stroma crosstalk at a global transcript level by the heterospecies heterospheroids model made up of the human pancreatic tumour cells Panc1 and the mouse PSCs (mPSCs) cultured under different serum/nutrient conditions. We discovered that Panc1 shifted from the classical to the basal-like subtype upon co-culture with mPSCs independent of serum condition. In addition, mPSCs acquired different CAF phenotypes upon co-culture with Panc1 under different serum conditions. Besides, mPSCs affected the chemosensitivity of Panc1 to different drugs, increasing the sensitivity to gemcitabine but decreasing sensitivity to paclitaxel and SN38. In paper III, we mainly investigated the function of CCN1, which had a significantly higher expression level in Panc1 from heterospheroids compared to monospheroids. Knockout of CCN1 by CRISPR-Cas9 technology in Panc1 resulted in elevated gemcitabine resistance, which was probably caused by the downregulation of gemcitabine transporting and metabolizing genes (SLC29A1 and DCK). In addition, we found that lysophosphatidic acid (LPA) and transforming growth factor β1 (TGFB1) signalling increased the expression of CCN1 in Panc1 cells. Besides, stimulation with LPA and TGFB1 also shifted mPSCs to a more myCAF-like phenotype, indicated by the increased expression level of Acta2 but decreased expression of Cxcl1. In conclusion, we developed and characterized a 3D co-culture model of human pancreatic cancer cells and mouse PSCs, which reflected key features of in vivo PDAC. Detecting interactions between tumour and stroma by the heterospheroids model showed novel roles of PSCs in promoting gemcitabine sensitivity of tumour cells, suggesting potential therapeutic opportunity in remodelling CAFs in PDAC.