Unbiased functional identification and therapeutic targeting of T cell neoantigens in a spontaneous murine squamous cell carcinoma [Bulk-seq]

The comparative resistance of some cancers including head and neck squamous cell carcinoma to checkpoint blockade is speculated to derive from the low frequency of expressed somatic mutations targeted by T cells as neoantigens. SCCVII, a spontaneously arising murine squamous carcinoma resembling human HNSCC in several key features, is likewise poorly immunogenic as irradiated tumor cells alone fail to induce protective immunity within syngeneic hosts. Justifying use of this model to identify NeoAgs, we confirm activated CD4+ and CD8+ T cells are detectable and essential for vaccine efficacy of SCC VII and polyI:C co-administration. Whole-exome sequencing tumor versus normal genome identified 39 nonsynonymous missense mutations that were synthesized into 81 representative 20-mers. NeoAg-specific CD4+ T cell IFN-γ responses were found against mutations of Pik3ca, Ctnnd1, and Otud5 while both CD4+ and CD8+ T cells produced IFN-γ when stimulated by a single Cltc mutation during in vitro recall assays. Prophylactic immunization with a mixture of all stimulatory peptides or the Cltc NeoAg alone protected hosts from subsequent tumor challenge. Further, the Cltc NeoAg, eliciting both CD4+ and CD8+ T cell responses, was also therapeutically beneficial in vivo. Anti-PD-1 combinatorial blockade resulted in synergistic tumor rejection via boosting Cltc-specific responses and increasing response diversification via epitope spreading. These data show that a functional NeoAg identification platform can be used to select immunotherapeutically relevant targets and filtration of neoepitopes that co-prime both CD4+ and CD8+ T cell responses is superior for practical intervention of poorly immunogenic tumors. For identificaion of NeoAg variants, which is described herein, sequence reads from Exome-Seq of the tumor and normal tail samples were aligned to the mm10 reference genome using SpeedSeq Align (v0.0.3a). Exome variants were identified using SpeedSeq Somatic (v0.0.3a) and variants were annotated using SNPeff (v4.1). Reads from RNA-Seq of the tumor were aligned using STAR aligner (v2.4.1c). In order to confirm the expression of exome variants in the tumor, the aligned RNA data was compared with the reference genome to check the presence of exome variants using Freebayes software (v0.9.16). Variants were considered for further analysis only if they met following criteria: sequencing depth at position of variant ≥ 10 in normal exome as well as in tumor exome and transcriptome, variant allele observations in normal exome = 0, variant allele observations in tumor exome ≥ 10, variant allele frequency in tumor RNA ≥ 20%, somatic score from SpeedSeq ≥ 0, SNP quality score ≥ 20, and the variant was protein coding and protein changing. For single nucleotide variants, two 20-mer peptides containing the mutation at positions 6 and 15 for each variant were generated using an in-house peptide generation pipeline. When the variant was near the start or the end of the protein sequence the mutation position was adjusted in the 20-mer peptide due to insufficient flanking amino acids. For frameshift variants, two long peptides each with mutated amino acid at positions 6 and 15 were generated in such a way that the peptide terminated at the first found stop codon in the mutated sequence. These long peptides were then broken into 20-mers with overlapping amino acids. Peptides used in experiments were synthesized by A & A Labs (DBA Synthetic Biomolecules, San Diego, CA).