Modelling melanoma control by immunotherapy and tissue-resident memory T cells using CRISPR/Cas9-based approaches

Abstract

In recent years, immunotherapy has demonstrated remarkable efficacy in the treatment of metastatic melanoma due to the development of T cell-based therapies such as checkpoint inhibitors or adoptive T cell transfer therapy (ACT) directed against defined antigens. However, tumours frequently relapse from therapy by diverse acquired resistance mechanisms. Currently, it is not well understood how the choice of target antigen influences resistance mechanisms to antigen-specific immunotherapies. A better understanding of tumour recognition by the immune system is of utmost importance to further improve currently used immunotherapies.<br /> Therefore, we established CRISPR-assisted insertion of epitopes (CRISPitope), a technique that fuses a model CD8⁺ T cell epitope, human gp100, to endogenous gene products. We applied CRISPitope to murine melanoma cells to tag the endogenous melanosomal protein, TYRP1, and the oncogenic protein, CDK4^R24C, with the same model epitope, rendering them targetable by the same TCR-transgenic T cells. This defined experimental setting enabled us to investigate how the choice of the targeted gene product impacts on therapy outcome and immune evasion mechanisms. <br /> Using experimental mouse models, we could identify different escape mechanisms to gp100-specific immunotherapy in TYRP1 versus CDK4^R24C melanomas. Resistance to ACT targeting TYRP1 was mainly caused by permanent antigen loss, accompanied by a non-inflamed microenvironment, or reversible downregulation of the antigen associated with melanoma phenotype switching. In contrast, CDK4^R24C melanomas escaping ACT displayed antigen persistence and were associated with an IFN-rich inflamed tumour microenvironment. In CDK4^R24C melanomas IFN-driven feedback inhibition by negative immune-checkpoint molecules promoted resistance to ACT despite persistent antigen expression. <br /> Applying CRISPitope to syngeneic mouse models, we could show that target antigen choice can influence ACT resistance mechanisms, phenotype and immune contexture of melanomas in response to antigen-specific immunotherapies. Thus, our work could help to better understand acquired resistance and optimise personalised cancer immunotherapy.<br /> Furthermore, we aimed to apply this platform to a model of melanoma immune surveillance by T_RM cells in order to understand the importance of cognate antigen expression and presentation for long-term tumour control by CD8⁺ tissue-resident memory T cells (T_RM). <br /> To address this question, we used a modified CRISPitope-approach, called SWITCHitope, to generate melanoma cell lines that express a floxed model antigen under the control of an endogenous promoter and a Tamoxifen-inducible Cre-recombinase.<br /> We could confirm successful Tamoxifen-inducible depletion of the model antigen in melanoma cells<em> in vitro</em> and<em> in vivo</em>. Moreover, we showed that antigen-depleted melanoma cells have significantly reduced potential to activate TCR-transgenic T cells <em>in vitro</em>. Using a transplantable epicutaneous melanoma inoculation technique, we could demonstrate that SWITCHitope-engineered melanoma cells can prime naïve T cells, recruit them into the skin and induce T cell differentiation towards a T_RM phenotype. <br /> Our approach enables us to investigate the importance of antigen expression and presentation for T_RM melanoma control. This work will help to better understand the interplay between tumour cells and T_RM cells and thereby advance clinical translation.