Cancer associated fibroblasts and T cells in pancreatic cancer : an immune perspective

Abstract

About 90% of pancreatic cancers are pancreatic ductal adenocarcinomas (PDAC) and its deadly hallmarks, including a prominent fibrotic stroma, low mutational burden, lack of early diagnosis and rapid metastasis, contribute to trial failures and to the poor five-year overall survival rate of only 9%. Carcinoma associated fibroblasts (CAFs) are the most abundant cell type in the stroma of PDAC and their interactions with tumor cells leading to tumor progression have been extensively studied. However, the role of CAFs in tumor immune evasion by impairing the functionality and migration of T cells is less understood. The main aim of this thesis was to isolate primary pancreatic CAFs to bring further knowledge on the CAF-T cell interactions in 2D and 3D in vitro models, as well as to study the phenotype and location of tumor infiltrating lymphocytes from resectable PDAC tumors.

Cultured primary pancreatic CAFs displayed an activated phenotype with a large variability between donors. CAFs expressed high levels of a-SMA, podoplanin, the PD-1 ligands PDL1 and PD-L2 and secreted large amounts of the pro-tumorigenic factors IL-6, periostin, and PGE2.In a non-contact dependent manner, CAFs inhibited T cell proliferation and activation, which led to T cell exhaustion and dysfunction by promoting the expression of the immune checkpoint receptors (PD-1, TIM-3, CTLA-4, LAG-3) on proliferating T cells and by inhibiting the secretion of effector cytokines (IFN-γ, TNF-α). T cells pre-cultured with CAFs lost the expression of CXCR3, known to promote T cell trafficking, and failed to migrate towards a CXCL10 gradient. Moreover, CAFs upregulated the expression of CXCR4, known to favor T cell exclusion from tumor nests. Blockade of CAF-released PGE2 partially restored the proliferation of T cells. Moreover, PGE2 suppressed T cell proliferation and upregulated the expression and co-expression of PD-1 and TIM-3 as well as CXCR4. Thus, blockade of PGE2 could be a promising target in combination with immune checkpoint therapies for PDAC.

Distinct CAF subsets with pro- and anti-tumorigenic functions have been identified in PDAC. In line with other studies, we found that the vitamin D3 analogue, calcipotriol, induced a less pro-tumorigenic phenotype by hindering the proliferative capacity of CAFs, and secretion of IL-6, periostin, PGE2, and LIF. However, CAFs treated or pretreated with calcipotriol had the same or even more pronounced suppressive effects on T cells than untreated CAFs. This was due to the strong suppression of calcipotriol on T cell proliferation and effector functions. The same results were observed in 3D models which resembled the tumor microenvironment rich in collagen I. Thus, even though vitamin D analogues could revert CAF phenotype to a less activated state it could also jeopardize the patients’ tumor immune surveillance.

Most of the tumor infiltrating lymphocytes in PDAC are not tumor reactive. However, infiltration of CD8+ T cells are associated with better overall survival (OS). We showed by immunohistochemistry stainings that very few T cells infiltrate the tumor nests, and that whereas CD4+ T cells were equally distributed throughout the stroma, CD8+ T cells were more distantly located from the tumor nests. Interestingly, a high ratio of CD8+ T cells close to the tumor nests relative to the total CD8+ T cells in the stroma was positively associated to the levels of CXCL9, CXCL10 and CXCL11, analyzed in the supernatants of consecutive tissues explants. This indicates a plausible role of these ligands on T cell migration within the pancreatic tumor microenvironment.

We also identified for the first time in PDAC, a recently described tumor specific T cell subset co-expressing CD39 and CD103 (DP). Flow cytometry analyses showed that DP CD8+ T cells accumulated in tumor rich areas compared to non-tumor tissues with strikingly high levels of PD-1 and TIM-3 and low basal levels of granzyme B suggesting an exhausted and dysfunctional phenotype. DP CD8+ T cells also presented a distinct chemokine receptor profile with high levels of CCR5 but low levels of CXCR3 and CXCR4. Kaplan Meier curves also showed that high co-expression of PD-1, TIM-3, and LAG-3 on CD8+ T cellsfrom tumor rich areas had a trend towards a worse OS, which highlights the importance of the immune checkpoints in pancreatic cancer.

Altogether, we showed that CAFs impaired T cell activation, effector functions and migratory capacity, which were not restored by calcipotriol but were partially mediated by PGE2. CD8+ T cells were mainly found distant from the tumor nests and their migration was partially regulated by CXCR3 ligands. We also found a small proportion of CD8+ T cells to be putatively tumor reactive which accumulated in tumor rich areas. Understanding the mechanisms by which CAFs suppress T cells and by learning more about the phenotype and functionality of tumor infiltrating T cells could lead to better combinational therapies for PDAC.