Exploring the impact of the PD-1/PD-L1 axis on CAR-T cell function

Abstract

[eng] Adoptive T cell transfer (ACT) therapy designed to express chimeric antigen receptors (CARs) has produced impressive clinical responses in certain cancer patients. Together with immune checkpoint blockade therapy, CAR-T cells are revolutionizing the field of cancer therapies. CARs are genetically engineered hybrid receptors that combine an antibody-derived extracellular domain with intracellular signalling domains derived from endogenous T cell receptors and costimulatory signals, which can induce T cell activation. The introduction of a CAR into a T cell successfully redirects the T cell with a new antigen specificity. Clinical outcomes achieved until date with CAR-T cell therapy for the treatment of solid tumours are yet far from the unprecedented success witnessed in hematologic malignancies. Despite this, recent works provide for the first- time clear evidence of objective antitumour responses in patients with hard-to-treat solid tumours. These results are highly encouraging and provide proof of the potential of CAR-T cells in this setting. Nevertheless, several obstacles remain to be addressed, including the trafficking of CAR-T cells to the solid mass, the tumour heterogeneity and loss of antigen expression, and the nutrient-restricted and immunosuppressive tumour microenvironment (TME), among others. One of the most prominent and well-studied T cell inhibitory axis is the programmed cell death protein-1 (PD-1)/ programmed death cell ligand-1 (PD-L1) checkpoint pathway. T cell activation following antigen recognition results in PD-1 upregulation, along with an intracellular signalling cascade that leads to the release of Th1 cytokines. These cytokines, in turn, induce the upregulation of inhibitory ligands such as PD-L1 on tumour cells but also on other cell populations within the TME. The interaction between PD-1 on T cells and PD-L1 on tumours ultimately leads to T cell suppression. As these activated T cells are potentially tumour specific infiltrating T cells (either endogenous or adoptively transferred T cells modified to express CARs), preventing the binding between PD-1 and PD-L1 might rescue antitumor T cell cytotoxicity and result in increased efficacy of cell-based

immunotherapies. In the context of CAR-T cell therapy, outcomes of disrupting PD-1 expression may vary depending on the specific CAR used, the type of tumour to be targeted or the different preclinical models employed (such as tumour cells engineered to express constitutively high levels of PD-L1, wild type tumour cell lines or patient derived cancer cells). The principal aim of this thesis was to examinate how different CAR configurations influence CAR-T cell sensitivity to PD-1/PD-L1 inhibition. Moreover, we sought to elucidate the impact of variations in target antigen density on CAR-T cell susceptibility to this inhibition. To address the challenge of model variability and deepen our comprehension of how different CAR constructs might be influenced by this pathway, we developed preclinical models expressing varying PD-L1 densities that better predict the efficacy of CAR-T cells. Our approach involved engineering tumour cell lines to express PD-L1 at absent, low, or high levels, enabling systematic investigation of diverse CAR configurations both in vitro and in vivo. Additionally, we established a synthetic model utilizing glass- supported lipid bilayers (SLBs) to precisely control the presence of target antigens and PD-L1 molecules, without additional inhibitors of CAR-T cell function. Through the utilization of these preclinical models, we delved into the impact of PD-1/PD-L1 axis inhibition on CAR-T cells designed to target specific antigens, with either low (LA) or high affinity (HA), and incorporating different co-stimulatory domains (i.e., CD28, ICOS or 4- 1BB). Our findings revealed that LA CAR-T cells exhibit heightened sensitivity to PD- 1/PD-L1 axis-mediated inhibition in comparison to HA CARs. Consequently, disruption of PD-1 enhanced the functional capabilities of LA CAR-T cells, while providing no discernible advantage to HA CAR-T cells. This trend was consistent across CARs featuring CD28 and ICOS co-stimulatory domains. Interestingly, CAR-T cells comprising 4-1BB co-stimulatory domain displayed intrinsic resistance to PD-L1-mediated inhibition. Furthermore, our observations suggest that low levels of the targeted antigen increased the susceptibility of HA CAR-T cells to inhibition via this axis.