Human tumor infiltrating lymphocytes cooperatively regulate prostate tumor growth in a humanized mouse model

Michael D Roth, Airi Harui

The complex interactions that occur between human tumors, tumor infiltrating lymphocytes (TIL) and the systemic immune system are likely to define critical factors in the host response to cancer. While conventional animal models have identified an array of potential anti-tumor therapies, mouse models often fail to translate into effective human treatments. Our goal is to establish a humanized tumor model as a more effective pre-clinical platform for understanding and manipulating TIL. The immune system in NOD/SCID/IL-2Rγnull (NSG) mice was reconstituted by the co-administration of human peripheral blood lymphocytes (PBL) or subsets (CD4+ or CD8+) and autologous human dendritic cells (DC), and animals simultaneously challenged by implanting human prostate cancer cells (PC3 line). Tumor growth was evaluated over time and the phenotype of recovered splenocytes and TIL characterized by flow cytometry and immunohistochemistry (IHC). Serum levels of circulating cytokines and chemokines were also assessed. A tumor-bearing huPBL-NSG model was established in which human leukocytes reconstituted secondary lymphoid organs and promoted the accumulation of TIL. These TIL exhibited a unique phenotype when compared to splenocytes with a predominance of CD8+ T cells that exhibited increased expression of CD69, CD56, and an effector memory phenotype. TIL from huPBL-NSG animals closely matched the features of TIL recovered from primary human prostate cancers. Human cytokines were readily detectible in the serum and exhibited a different profile in animals implanted with PBL alone, tumor alone, and those reconstituted with both. Immune reconstitution slowed but could not eliminate tumor growth and this effect required the presence of CD4+ T cell help. Simultaneous implantation of human PBL, DC and tumor results in a huPBL-NSG model that recapitulates the development of human TIL and allows an assessment of tumor and immune system interaction that cannot be carried out in humans. Furthermore, the capacity to manipulate individual features and cell populations provides an opportunity for hypothesis testing and outcome monitoring in a humanized system that may be more relevant than conventional mouse models.