A pan-allele Class I MHC prediction tool in mice enables tumor immunology studies

Abstract

Abstract An understanding of tumor-presented Class I MHC peptides is a prerequisite to effective cancer immunotherapy, and more tools are needed to elucidate immune interactions in breast cancer. One tool, the mouse mammary tumor virus (MMTV)-Polyoma Middle T Antigen (PyMT) mouse, models the development of spontaneous metastatic breast cancer in vivo. Surprisingly, little data exists for the MHCI of MMTV-PyMT tumors, which express the H2-q haplotype. To develop the MMTV-PyMT tumor model for tumor immunology, we present an open-access Class I MHC peptide prediction tool for multiple murine haplotypes (including H2-q): NetH2pan (http://www.cbs.dtu.dk/services//NetH2pan/). H2-d/k/b peptides were obtained from the IEDB, while H2-q peptides were eluted from MHC: HeLa and 721.221 cell lines transfected with PyMT and soluble H2-Kq/Dq. Cells secreting these MHC I were expanded, peptide-bearing MHC I affinity purified, and presented peptides analyzed by mass spectrometry. The identified peptide sequence data were used to develop a predictive algorithm to enable in silico candidate peptide searches. Validation of the algorithm using peptides eluted from tumors demonstrated a positive predictive value (34%) improved from other MHC I prediction tools. For a candidate antigen PyMT, the 5 PyMT-peptides eluted from tumors were in the top 25 predicted. Three of 5 peptides were immunogenic as determined by IFNγ ELISpot and peptide:MHCI tetramers in vaccinated mice, and correlated with tumor prevention. These data show that NetH2pan can predict immunogenic epitopes with high-fidelity and that these peptides can be utilized for vaccine development. These studies enable the MMTV-PyMT model for mechanistic studies involving tumor-T cell crosstalk.