Background T cells that target neoantigens, which are mutant peptides encoded by tumor-specific mutations in cancer cells, have a pivotal role in response to immunotherapy. Current affordable NGS and bioinformatics have facilitated neoantigen research. However, the prediction of neoantigens in silico analysis lacks actual proof that the predicted neoantigens are indeed relevant to anti-tumor immune response. In this study, immunogenicity and anti-tumor activity of in silico predicted neoepitopes of gastric cancer cell lines, YTN2 and YTN16, were evaluated in vitro and in vivo. Besides, MHC ligandome profiling was performed on whether predicted neoantigens were presented by cancer cells.

Methods YTN2 and YTN16 are subclones established from a chemically induced murine gastric cancer (Cancer Sci. 2018;109:1480-1492). Whole-exome sequencing and microarray analysis were performed. Binding affinities of peptides with MHC-I were predicted using NetMHCpan. Peptides with IC50≤250 nM were considered as candidate neoantigen.

Results Exome sequencing identified 3329 and 3347 missense mutations in YTN2 and YTN16 tumors, respectively. Most of the mutations (3007) were common in YTN2 and YTN16. Among them, 2356 and 2166 mutations were expressed in YTN2 and YTN16, respectively. Candidate neoantigen peptides were 263 and 250 in YTN2 and YTN16, respectively. Although we detected 75 neoantigen-peptide-specific T cell responses, only six neoantigens (Cdt1, Cers4, Nlrp1a, Scarb2, Vmn2r121, and Zfp106)-specific T cells could recognize YTN2 or YTN16 cell line and produced IFNγ. To evaluate in vivo anti-tumor effects of neoantigen-specific T cells, we expanded four neoantigens (Cdt1, Nlrp1a, Vmn2r121, and Zfp106)-specific CD8+ T cell lines and injected them into YTN16 tumor-bearing mice. Only Cdt1-specific T cells completely eradicated the tumor, and Nlrp1a-specific T cells partially inhibited tumor growth. However, Vmn2r121- or Zfp106-specific CD8+ T cell lines could not inhibit the tumor growth. MHC ligandome profiling identified 424 Db-binding peptides and 228 Kb-binding peptides in YTN2, while 317 Db-binding peptides and 228 Kb-binding peptides were detected in YTN16. Almost all MHC-binding peptides had wild type sequences. Only one mutated peptide, Cdt1, was detected in YTN2 but not YTN16. Although Cdt1-specific CD8+ T cells recognized both YTN2 and YTN16. Other neoantigens, Cers4, Nlrp1a, Scarb2, Vmn2r121, and Zfp106, were not detected by MHC ligandome profiling, though T cells specific to these neoantigens responded to cancer cells.

Conclusion Currently available technologies anyhow identified neoantigens that are relevant to anti-tumor immune response. However, any single methodology is not sufficient to identify real neoantigens. Detection of neoantigen-specific T cells is still important.

Citation Format: Koji Nagaoka, Takayuki Kanaseki, Changbo Sun, Akihiro Hosoi, Serina Tokita, Kazuhiro Maejima, Masashi Fujita, Hidewaki Nakagawa, Toshihiko Torigoe, Kazuhiro Kakimi. To identify neoantigens relevant to anti-tumor activity is challenging even with NGS-based prediction combined with MHC ligandome profiling [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 6610.