O15.05 Development of peptide-centric CARs targeting IGFBPL-1 for cancer immunotherapy in small-cell lung cancer

Background Targeting intracellular antigens has to date been impractical for most immunotherapies, due in part to limited accessibility and difficulty identifying reactive T cell receptors (TCRs). However, artificial intelligence (AI) derived de novo peptide design has challenged this paradigm, for the first-time allowing rational design of binders specific for peptide-major histocompatibility complexes (pMHC), which can be utilised as antigen-binding moiety in peptide centric CAR T cells. 1 Material and Methods To exploit these new capabilities, we have identified an HLA-A*02 restricted peptide derived from IGFBPL-1 (Insulin-Like Growth Factor-Binding Protein 1) frequently expressed in SCLC (small cell lung cancer). Initial pMHC binder candidate designs were generated using RFdiffusion, 2 followed by a custom-made iterative refinement algorithm. To experimentally identify specific binders, we implemented a high-throughput library screening strategy utilising mammalian cell surface expression. Positive cells were sorted via HLA-A*02-IGFBPL-1 tetramer staining, with enriched binders identified via next-generation sequencing. These screening hits were incorporated as binding moieties of peptide-centric CARs for functional validation in primary human T cells.Results We achieved an initial binder design success rate of