Dissection of the T-cell infiltrate in mouse pancreatic tumors reveals an extensive and diverse tumor-reactive T-cell repertoire

Background: Pancreatic ductal adenocarcinoma (PDAC) is refractory to cytostatic and immunotherapeutic treatment regimens, including immune checkpoint inhibition (ICI). The latter is due to the limited mutational load of most of these tumors, in combination with the immunosuppressive hurdles imposed by the tumor microenvironment. Here we present an in-depth analysis of the T-cell response at the single-cell level in an orthotopic PDAC mouse model that recapitulates the therapy resistance of the human disease. In parallel, we investigated an ICI-sensitive variant of this model expressing an immunogenic neo-antigen. Methods: Single-cell RNA and T-cell receptor (TCR) from tumor-infiltrating lymphocytes (TILs) of 14 tumor-bearing mice were combined with functional in vitro screening of TCR specificity against tumor. Transcriptomic differences between tumor-reactive (TR) and non-tumor-reactive (NTR) T-cells were harnessed to define gene signatures for the purpose of predicting the specificity of additional TCRs. Methods: We used scRNA and scTCR sequencing to analyze tumor-infiltrating T-cells in a model of murine PDAC for tumor-reactivity. Testing of T-cell receptors in vitro against tumor cells revealed specific anti-tumor activity of the majority of T-cells in the PDA-OVA model, and to a lesser but significant extent in the less immunogenic PDA30364 model. Localization of tested T-cells on the transcriptome revealed distinct profiles of tumor-specific T-cells, dominated by an exhausted phenotype, and of bystander T-cells, which displayed a T effector memory phenotype. Underlying differences in gene expression were delineated for generation of gene signatures that confidently discriminated tumor-reactive and bystander T-cells in this PDAC model, as well as in the CRC model MC38. Results: Our efforts revealed extensive tumor-reactive and bystander CD8+ and CD4+ T-cell repertoires, not only in the immunogenic but also in the ICI-refractory tumors. By integrating the functional screening and single-cell transcriptome data sets, we delineated gene signatures that distinguish between tumor-reactive and bystander T-cells in the mouse model as well as in human cancers. The TCRs in the murine models detect a range of tumor-associated antigens beyond mutanome-encoded neoepitopes and retain functional anti-tumor activity in vitro. Overall design: Transplantable tumor cell lines PDA30364 and PDA-OVA, which are derived from a genetically engineered mouse model (Elas-tTA/TetO-Cre Kras+/LSL-G12D Trp53+/LSL-R172H), were implanted orthotopically into the pancreas of immune-competent C57BL/6 mice. Mice were subjected to control treatment or treatment with MEK inhibitor GDC-0623 in combination with agonist anti-CD40 antibody treatment. After two cycles of treatment, tumors were harvested from mice, dissociated into single-cells for the sorting of tumor-infiltrating T-cells. T-cells were analyzed for single-cell RNA and VDJ expression using 10xGenomics for discrimination of tumor-reactive T-cells from bystander T-cells. Transplantable tumor cell line MC38 was implanted subcutaneously in immune-competent C57BL/6 mice. Tumors were harvested from untreated mice, dissociated into single-cells for the sorting of tumor-infiltrating T-cells. T-cells were analyzed for single-cell RNA and VDJ expression using 10xGenomics for discrimination of tumor-reactive T-cells from bystander T-cells based on the gene expression signatures derived from the PDAC tumor-infiltrating T-cells.