(P 333) ESTIMATING IMMUNE INFILTRATES OF NOVEL GENETICALLY ENGINEERED MOUSE MODELS OF UNDIFFERENTIATED PLEOMORPHIC SARCOMA

Objective: Undifferentiated Pleomorphic Sarcoma (UPS) is the most common soft-tissue sarcoma in older adults. Treatment options remain limited, as UPS tumors do not respond to many systemic therapies and immunotherapies. Genetically engineered mouse models of UPS provide an opportunity to test the efficacy and mechanism of novel therapeutics but vary in their ability to recapitulate the spectrum of human disease states, including tumor and immune cell types and states. Here, we characterize immune infiltration from UPS mouse models which are driven by combinations of Pten, Trp53 and Kras mutations. The orthologous human genes PTEN and TP53 are frequently altered in UPS and the related cell cycle, PI3K, and TP53 pathways are the most frequently mutated pathways across most soft tissue sarcomas. We further assess the impact of reporter gene expression and the difference between spontaneous and cell-line-derived tumors for each of the driver gene combinations.

Methods: The models and the driver genes included in the study are KPP (Trp53fl/fl; Ptenfl/fl; KrasG12D/+), PTEN (Trp53fl/fl; Ptenfl/fl), and KP (Trp53fl/fl; KrasG12D/+). For each of these models spontaneously induced tumors (spKPP, spKP and spPTEN) and corresponding cell line engrafted tumors were generated. For each of the three spontaneous models, cell line engrafted tumors were generated with (+) and without (-) a mCherry-Luciferase reporter construct (KPP1053+ and KPP1053-, TAO+ and TAO-, UP10+ and UP10-, respectively) for a total of 9 models and 5 replicate tumors per model. Bulk RNA sequencing data from a total of 45 mice were analyzed to identify differentially expressed genes (DESeq2) and to estimate the proportions of tumor infiltrating immune cells using combination of supervised immune deconvolution methods (immunedeconv). We then compared the spontaneous models to each other, spontaneous tumors and their corresponding engrafted tumors (without the mCherry-Luciferase reporter), and engrafted tumors with and without the mCherry-Luciferase reporter.

Results: Compared to spKP and spPTEN tumors, spKPP resulted in a large upregulation of genes associated with immune response pathways. spKP resulted in enrichment of Hippo signaling, Wnt signalling and TGB-signaling pathways when compared to spKPP, and pathways related to morphogenesis, cell-cell adhesion, response to growth factor and developmental growth when compared to spPTEN. The genes upregulated in spPTEN were associated with tissue morphogenesis and development, extracellular matrix organization, anatomical structure homeostasis, cell migration and pathways in cancer. Immune deconvolution results showed that among the spontaneous models spKPP has the highest immune proportions of mast cells and T cells while endothelial cells were highest in spPTEN. We compared the spontaneous models with their corresponding cell line engrafted models and found the differentially expressed genes were enriched in immune response pathways in spKP and spKPP. These pathways included innate and adaptive immune response, cytokine production, T cell activation, complement cascade and cellular response to interferon-beta among others. Immune deconvolution revealed spKPP to have higher proportions of B cells, T cells, macrophages, monocytes, NK cells and myeloid cells while spKP had higher gamma delta T cell, CD8 effector cells, CD8 memory cells and NK cells as compared to TAO- cell line tumors. The spPTEN tumors did not show a marked enrichment in immune regulatory pathways and in immune cells as compared to corresponding cell line models. Lastly in the comparisons within cell line engrafted tumors, the differentially expressed genes in KPP1053+ and TAO1+ are enriched in immune response regulation, lymphocyte activation, cytokine production, NK cell mediated immunity, defense response to virus and B-cell activation among others. The proportion of immune cells revealed higher infiltration in models with mCherry-Luciferase reporter construct, suggesting that the reporter protein is immunogenic.

Conclusion: In this work, we characterize three models of UPS with different driver gene combinations and assess their impact on gene expression and immune infiltration based on RNA-Seq data. We show that driver genes influence immune cell infiltration. We further demonstrate that cell line engrafted tumors have highly consistent immune infiltrates across replicates, and that engrafted tumors tend to have lower immune cell infiltration than spontaneous tumors or those that also express the reporter construct especially in the KP and KPP models. These models represent a resource for translational research studies to propel the discovery of novel treatments for UPS.