Simultaneous high-throughput identification of myeloid leukemia driver mutation-derived public neoantigens and cognate functional TCRs for TCR-T therapy

Abstract Many myeloid malignancies are driven by acquisition of recurrent somatic mutations which increase cancer cell fitness and growth. These “driver mutations” have the potential to produce public neoantigens, which are attractive immunotherapy targets. To produce a neoantigen, a mutation-containing peptide must be processed from a full-length protein by the proteasome, loaded onto an HLA molecule, and presented on the surface of the cell. However, identification of neoantigen-containing peptide:HLA (pHLA) is challenging because most mutation-containing peptides are not presented. While mass spectrometry/immunopeptidomics can be used to identify cell surface HLA-presented peptides, false negatives are common due to failure to detect low abundance (but immunologically relevant) pHLAs. In comparison, T cells are extremely sensitive; they are capable of detecting and responding to a single copy of a pHLA on a target cell. Here we describe a technique that leverages the exquisite antigen sensitivity of T cells to systematically and simultaneously identify immunogenic driver neoantigen-derived pHLAs, and their cognate TCRs. Driver mutation-containing mRNA tandem minigene constructs (TMG) were designed to encode all possible mutation-spanning 8-11mer peptides from 49 common cancer driver mutations found in AML and other malignancies. TMG mRNA was transfected into monocyte-derived dendritic cells (moDC), which were then co-cultured with autologous naïve CD8 T cells to expand T cells with neoantigen pHLA specific T cell receptors (TCRs). In this system, driver mutation-derived pHLAs stimulate T cells only if they are processed by the proteosome, loaded onto HLA, and presented on the surface of the DC. In parallel, all possible neoantigen-containing 8-11mer peptides for the 49 driver mutations were synthesized and tested for HLA binding using a UV-mediated ligand exchange-based ELISA. In total, 15,704 unique pHLA complexes across 8 common HLA class I alleles were screened. Eight percent of peptides were found to bind an HLA molecule, 89% of which were not predicted to bind by HLA binding algorithms. Subsequently, a pooled DNA-barcoded pHLA-tetramer library (containing all identified binder pHLAs), was constructed, and used to stain TMG-driven expanded T cells. Tetramer-binding cells were sorted and sequenced using 10X Genomics kit-based single cell workflows to retrieve the paired TCRab sequences and their antigen specificity. We identified expanded, polyclonal T cell responses against 29 known and novel neoantigen pHLA in myeloid malignancy (including those containing FLT3:D835Y, IDH2:R140Q, DNMT3A:R882H/C, CALR-frameshift, KIT:D816V, TP53:R175H, and KRAS:G12V/R, among others).To validate our findings, TCR-T cells expressing the novel TCRs were generated and tested functionally. TCR-T cells bound to the mapped neoantigens via fluorescent tetramer staining and exhibited dose-dependent activation against neopeptide-pulsed B-LCL. We further confirmed that the epitopes identified are bona fide cancer neoantigens, as we observed antigen-specific reactivity and cytotoxicity of TCR-T cells against cancer cells that endogenously express the mutations. Some neoepitopes may be post-translationally modified in cancer cells which can hinder TCR recognition, for example lysine methylation(s) on KRAS:G12V-derived peptide presented on HLA-A2. Our system identified TCRs that reacted to both native and methylated neoepitopes of KRAS:G12V as well as cancer cells that endogenously express the mutation. Finally, engineered TCR-T cells specific to DNMT3A:R882H on HLA-A*01:01 or IDH2:R140Q on HLA-B*07:02 demonstrated selective killing of primary human AML cells harboring the mutations and HLA molecules, demonstrating that these peptide-HLAs are presented on leukemia cell surface and are targetable by T cells. Together, we developed a sensitive and high-throughput approach that detects novel public neoantigens derived from cancer driver mutations at scale and allows simultaneous retrieval of their cognate TCR sequences for potential clinical applications.