Abstract:21 Cancer mutations lead to the production of mutated proteins from which can be processed and 22 presented to CD8 T cells as neoantigens. Despite the promise of immunotherapy targeting 23 neoantigens, which mutant peptides are likely to induce immune responses remain elusive. Here, we 24 determined two sets of MHC-I bound neoantigen structures and performed bioinformatics investigation 25 based on neoantigen data. Structural and bioinformatic analyses reveal that some types of mutations in 26 immunogenic neoantigens are preferentially selected. We also developed a strategy to expand the usage 27 of a known neoantigen by scanning potential MHC alleles. Overall, our results will extend the ability to 28 discover neoantigens that may be useful for cancer immunotherapy. 29 30 Introduction 31 The immune system can sometimes recognize and destroy cancer cells(1). CD8 T cells have been 32 demonstrated to be a major component in this process by both recognizing and destroying the target 33 cells(2–5). For T cells to react with tumor cells two components are required, major histocompatibility 34 complex class I and possibly class II proteins (MHC-I, MHC-II) on the tumor cells and peptides which 35 processed by the tumors bound to these MHC proteins. CD8 T cells can then recognize the 36 combination of MHCI and tumor derived peptides on the cancer cells. Thus, it is crucial to identify 37 effective tumor antigenic peptides with their relevant MHCs. 38 During the last 25 years, great efforts have been made to identify tumor antigens(6). These tumor 39 antigens can be classified into two broad categories, self-antigens which are seldom expressed in 40 normal tissue or expressed at much higher levels on the cancer cell versus normal tissues (like NY41 ESO-1, MART-1) and non-self-antigens called neoantigens which are derived from somatic mutations 42 in the tumor(7). The neoantigens are not presented by normal tissues, hence the immune system is not 43 tolerant to them and views them, correctly, as foreign antigens and responds appropriately(8, 9). 44 Recently, next-generation sequencing (NGS) provide formidable identification of tumor-specific non45 synonymous mutations, frameshift mutations, and gene rearrangements, which have driven clinical 46 immunotherapeutic methods such as vaccination, adoptive cell transfer (ACT) and blockade of the 47 proteins that inhibit T cell activation, have demonstrated that immunotherapy targeting tumor 48 neoantigens can induce objective clinical responses in many patients with metastatic cancers(10–13). It 49 appears that T cell targeting of tumor neoantigens might represent the “final common pathway” to treat 50 cancer(14). However, neoantigen discovery and validation remains a daunting problem. Classical 51 cDNA expressing screening methods to identify tumor neoantigens without knowledge of the 52 properties needed for a mutated peptide to be a successful immunotherapeutic is not efficient(15). To 53 solve this problem, recently, a study identified mouse neoantigens by combination of NGS and 54 prediction algorithms for antigen presentation by MHCI(16). In brief, they compared mouse tumor and 55 normal DNA for identifying non-synonymous mutations and used peptide/MHCI binding prediction 56 algorithms to find mutated peptides which could bind MHC to be potential effective neoantigens. 57 These advances of discovering neoantigens bring us a step closer to personalize immunotherapy. 58 Indeed, neoantigen prediction lighten the burden of immunogenicity testing by reducing the number of 59 candidate peptides(17–27). Nevertheless these algorithms are not robust enough to identify 60 immunogenic neoantigens accurately(14, 28). There are several reasons for this. Firstly, many of these 61 algorithms do not consider the likelihood that the peptide in question will be properly processed and 62 transported to the endoplasmic reticulum, where they bind MHCI. This problem can be solved by 63 adding other methods, such as NetChop(29) (for antigen processing prediction) and NetCTL(30) (for 64 peptide transport prediction) to the prediction process. Secondly, the MHC binding prediction methods 65 based on artificial neural network (ANN) models need a lot of data ,such as those available in NetMHC 66 4.0, for training(31). However, large amounts of these input data are lacking for many human MHC 67 alleles. Also, in some cases, most peptides in the training data were of non-tumor origin such as 68 cultured cell lines and microbial foreign peptides. Thus, it is unclear whether these non-self peptidome 69 data accurately represent the profile of mutated-self neoantigens and elicit T cell response(32). Thirdly, 70 among neoantigens binding MHC actually, only a small portion of them are immunogenic and have 71 therapeutic effects(33). Thus, it is crucial to investigate the details of neoantigen immunogenicity and 72 to improve the performance of in silico neoantigen prediction algorithm. 73 Recently, Tran et al. reported a remarkable case of a patient whose metastatic colorectal cancer 74 regressed after infusion of HLA-C*08:02-restricted tumor-infiltrating lymphocytes (TILs) targeting 75 KRAS G12D neoantigens. Kras encodes a small GTPase that mediates downstream signaling from 76 growth factor receptors, and it is one of the most infamous driver genes led to oncogenesis. The most 77 common KRAS mutation is a gain-of-function substitution of the glycine to aspartic acid at codon 12 78 (KRAS G12D), a change that is often present in colorectal or pancreatic cancer(34–36). The KRAS 79 G12D mutant 9mer peptide GADGVGKSA and 10mer peptide GADGVGKSAL can both stimulate 80 autologous T cells in the context of HLA-C*08:02, while their wild type counterparts cannot(23). 81 Adoptive cell transfer (ACT) of T cells specific for the single mutant neoantigen led to tumor 82 disappearance in patient. Thus, the KRAS G12D neoantigens have been acknowledged to be useful and 83 typical neoantigen that might be used as a model to illustrate the secret of therapeutic tumor 84 neoantigens. 85 Here we took structural analyses of several neoantigens with their relevant MHCs, combined with a 86 broad examination of immunotherapeutic and non-immunogenic tumor antigens. We concluded that a 87 universal type of tumor neoantigen peptide mutations from non-preferential residues to preferential 88 residues with their relevant MHCs at anchor positions is unexpectedly a good indicator to predict 89 neoantigen immunogenicity. Combined with other findings about effective neoantigens we thus 90 propose an optimized strategy for neoantigen identification. 91 92

The Workflow of Neoantigen Identification Based on Proteogenomic Methodology

ProGeo-neo: a Customized Proteogenomic Workflow for Neoantigen Prediction and Selection

Proteogenomics offers a novel avenue in neoantigen identification for cancer immunotherapy

A proteogenomic approach to target neoantigens in solid tumors

Proteogenomics guided identification of functional neoantigens in non-small cell lung cancer

Application of Mass Spectrometry-Based MHC Immunopeptidome Profiling in Neoantigen Identification for Tumor Immunotherapy

A comprehensive proteogenomic pipeline for neoantigen discovery to advance personalized cancer immunotherapy

A Novel Proteogenomic Integration Strategy Expands the Breadth of Neo-Epitope Sources

Advances in the Prediction of Antigenic Peptides in Personalized Tumor Neoantigen Vaccine

Cancer Neoantigen Prioritization Through Sensitive and Reliable Proteogenomics Analysis

TSNAdb: A Database for Tumor-specific Neoantigens from Immunogenomics Data Analysis.

TruNeo: an integrated pipeline improves personalized true tumor neoantigen identification

312P Identification of Neoantigen-Specific T Cell Response and Anti-Tumour Immunity in Pancreatic Cancer

Key Parameters of Tumor Epitope Immunogenicity Revealed Through a Consortium Approach Improve Neoantigen Prediction

PGNneo: A Proteogenomics-Based Neoantigen Prediction Pipeline in Noncoding Regions.

Characters of Neoantigens in Cancer Immunotherapy 1

Novel Pipeline of High-Frequency Neoantigens Heathy Donor-Based Validation in Breast Cancer

ProGeo-Neo v2.0: A One-Stop Software for Neoantigen Prediction and Filtering Based on the Proteogenomics Strategy

Toward in silico Identification of Tumor Neoantigens in Immunotherapy

Neoantigen Identification Strategies Enable Personalized Immunotherapy in Refractory Solid Tumors

A functional identification platform reveals frequent, spontaneous neoantigen-specific T cell responses in patients with cancer