Functional characterization of neoantigens determining immune checkpoint blockade response in mouse models of human melanoma

Melanoma is the deadliest form of skin cancer due to the lack of widely effective therapies for advanced disease. Recently FDA-approved immunotherapies, such as immune checkpoint blockade (ICB) by CTLA-4 and PD-1/PD-L1 antibodies, provide unprecedent durable responses but in less than 40% of late stage melanoma patients. While high mutational loads characteristic of responsive tumors has not shown predictive value of patient outcome, accumulating evidences suggest a key role for neoantigens in the response to ICB. Moreover, new T-cell transfer- and vaccine-based therapeutic strategies highlighted the relevance of an efficient identification and prioritization of highly immunogenic neoantigens to improve immunotherapies. However, mechanistic studies are not possible in humans, and the development of adequate predictive methods is still the center of intense debate in the field. Here we use two genetically engineered melanoma mouse models exhibiting distinct response to ICB. Melanomas induced by neonatal ultraviolet radiation (UV) in a HGF-transgenic mouse (HGF-tg) showed high sensitivity to anti-CTLA-4, whereas UV-induced melanomas in HGF-tg; Braf V600E ; Cdkn2a +/- mouse (Braf/HGF) did not respond. We hypothesized these models will allow us to identify the neoantigen features required for ICB response including type, expression levels and allele frequency patterns. High tumor immunogenicity, assessed by in vivo vaccination assays, as well as increased T-cell infiltration upon anti-CTLA-4 treatment were correlated to greater response. Moreover, exome and RNA sequencing analyses revealed similar mutational and neoantigen load in both models, albeit with no common expressed mutations. Notably, both models expressed similar levels of antigen presentation related genes (e.g. B2m, H2-Kd, Tap1) suggesting that specific neoantigens in HGF-tg melanoma cells may contribute to their sensitivity to anti-CTLA-4. To test this, we predicted MHC-I/-II binding of HGF-tg melanoma mutated epitopes in silico and generated a “neo-epitope” library. Importantly, the expression and allele frequency of most of selected mutations were decreased in anti-CTLA-4 responder HGF melanomas. The “neo-epitope” library was transduced into Braf/HGF non-responder cells and future studies will identify the neo-epitopes lost upon ICB, representing determinants of therapeutic success. Additionally, in vivo vaccination assays using synthetic mutant peptides will be performed to validate each neoantigen candidate. We anticipate that our studies will provide insight into the role that neoantigens play in melanoma immunotherapy responses. (EPG and CPD contribute to this abstract equally). Citation Format: Eva Perez Guijarro, Chi-Ping Day, Zoe W. Ohler, Rajaa El Meskini, Howard Yang, Suman Vodnala, Cari Graff-Cherry, Sung Chin, Anyen Fon, Helen Michael, Maxwell Lee, Terry Van Dyke, Shyam Sharan, Glenn Merlino. Functional characterization of neoantigens determining immune checkpoint blockade response in mouse models of human melanoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5720.