Genetic screens to improve immunotherapy

In this thesis, I describe our efforts to understand immune resistance mechanisms, frequently by the use of CRISPR-Cas9, functional genetic screens. I approach this problem from two sides: tumor-intrinsic and tumor-extrinsic immune resistance mechanisms. I introduce these concepts in Chapter 1. In Chapter 2, Julia and I describe the discovery of a novel, multi-therapy resistant subpopulation of melanoma. This neural-crest like population, marked by the expression of NGFR, is present within treatment-naïve melanoma and expands upon (immuno)therapy. Intriguingly, and extending earlier observations, this NGFR+ subpopulation is stable and can be maintained during extended in vitro culture and in in vivo experiments. In patients, this population can also be found and its presence is correlated with non-response to ICB. This resistance in patients may be caused by active T cell exclusion, since we found a clear absence of T cells within NGFR+ subregions of heterogeneous tumors. In Chapter 3, I present work by Thomas and myself, where we present a novel strategy to enhance immunotherapy efficacy. By performing a CRISPR-Cas9 screen we found that the loss of a number of genes within the TNF signaling pathway could sensitize tumor cells to T cell attack. Particularly the loss of TRAF2 or its binding partner BIRC2, could sensitize tumor cells in vitro, and murine tumors in vivo to T cell-derived TNF to which they are normally poorly susceptible. This matched our clinical analyses, in which we could only find evidence for an antitumor role of TNF in patient tumors responding to ICB. A combined TRAF2/cIAP targeting strategy, potentially through agonistic engagement of the TWEAK receptor, may be a matching therapeutic strategy. Together with Georgi, I present our identification and characterization of STUB1 as a regulator of cell-surface expression of IFNγ-R1 in Chapter 4. Firstly, we show that the level of IFNγ-R1 in tumor cells is directly correlated with the extent of response to IFNγ, and their susceptibility to T cell attack. Then, through a sort-based, genetic CRISPR-Cas9 knockout screen, we identified STUB1 as a major regulator of cell-surface IFNγ-R1 expression. In extension to the tumor-intrinsic work described above, Georgi and I present an overview and meta-analysis of the results obtained through tumor-intrinsic CRISPR-Cas9 screens for immune sensitizers in Chapter 5. In this work, we establish that there is significant overlap between the different screens, identifying immune sensitivity modifiers in the TNF, IFNγ, antigen presentation and autophagy pathways, but that there are also curious contradictions. These can be found in the exact node within a particular signaling pathway that is identified and the different performance of those nodes within different experimental settings. In Chapter 6, I present my CRISPR-Cas9, sort-based screen for PD-1 regulators in primary murine T cells. We found that the TMED family of proteins, most prominently TMED10, regulates the level of PD-1 expressed by CD8 T cells. Mechanistically, TMED10 interacts with PD-1, and promotes its delivery to the cell surface. Genetic perturbation of TMED10 limits this delivery and promotes T cell function. Treatment of T cells with AGN192403, a pan-TMED inhibitor, phenotypically recapitulates the genetic loss of TMED10. Additionally, in vivo, this inhibitor promotes both the fitness of CD8 T cells, by overcoming T cell dysfunction, and their antitumor efficacy. In patients, this can be recapitulated, as in patient tumor infiltrating lymphocytes (TIL), TMED expression is correlated with T cell dysfunction and TMED expression in CD8 T cells can be used to predict immunotherapy response. In Chapter 7, I present my perspective on the past and future of tumor-intrinsic genetic screens, as I feel that heterogeneity is imperfectly captured within the screens performed to date, including our own.