Abstract 4891: Identification of recurrent high-affinity MHC class I restricted neo-epitopes in neuroblastoma using ProTECT

Abstract

Abstract Introduction: T-cells are trained to differentiate between cell-surface MHC-displayed peptide sequences from self- and non-self proteins and act on the latter. The numerous mutations often associated with cancers can occur in coding regions of the genome and modify the sequence of wild-type proteins, potentially creating targets for immunotherapies. We have developed an analysis pipeline ProTECT (Prediction of T-cell Epitopes for Cancer Therapy) to identify and rank neo-epitopes in terms of immunogenicity. Running ProTECT on a set of Neuroblastomas patients predicted hotspot mutations that bind well to high-frequency MHC alleles – combinations that would potentially benefit a large subset of NBL patients. Methods: ProTECT accepts paired tumor and normal DNA sequencing fastq files, and tumor RNA sequencing fastqs. Mutations are called using a panel of callers [1-4], and are annotated[5] to identify coding mutations. Prediction of self-MHC:mutated-peptides is carried out[6] and the final binding predictions are ranked using an in-house algorithm. Summary: Running ProTECT on 6 Neuroblastoma samples (NBL) from the TARGET (Therapeutically Applicable Research to Generate Effective Treatments) project revealed 2 well-known hotspot mutations in NBL (NRAS Q61K and ALK R1275Q) that bind to common MHC alleles (A*01:01 and B*15:01 respectively). We also found 2 closely related mutations in ALK F1174L and F1174I that are predicted to C*04:01 and C*07:02. We carried out in-vitro refolding and crystallization assays [7] for the five highest-ranking mutant NRAS and ALK R1275Q predictions. Properly conformed MHC trimers were verified by a monodisperse peak after anion exchange chromatography. SDS gel electrophoresis and Mass-spec confirmed bound peptide for 4/5 tested predictions and 3 of these were used to set up hanging-drop crystallization trials in various conditions. Positive hits were obtained for one (AQDIYRASY::HLA-B*15:01) and the structure was obtained at 1.7A. The structure suggested the binding of the 10-mer (AQDIYRASYY) to the MHC and this was shown to bind better than the 9-mer using Differential Scanning Fluorimetry[9]. We will run ProTECT on the remaining 100+ TARGET NBL trios, and on relevant cohorts within The Cancer Genome Atlas (TCGA). We aim to reveal clinically relevant hotspot-mutation:MHC pairs. Conclusion: We have described a pipeline for identification and ranking of therapeutically relevant neo-epitopes. We have predicted potential targets for NBL that have been validated in-vitro.