Abstract 1088: Exploration of the human peptidome for cancer-specific peptide-MHC targets

Abstract

Abstract Introduction MHC class-I molecules display peptide fragments of intracellular proteins on the cell surface. Peptide-MHC (pMHC) complexes containing peptide fragments derived from antigenic proteins lead to a T-cell mediated immune response. We are interested in cancer-specific pMHC complexes that are targetable by therapeutic molecules, e.g., bispecific antibodies. A significant concern is off-target toxicity of the therapeutic molecules due to pMHC complexes similar to the target complex in essential, normal tissues. Thus we performed a comprehensive exploration of the 9-mer human intracellular peptidome to computationally identify potential breast cancer-specific pMHC class I targets and evaluate the likelihood of cross-reactivity. Methods Our method has four steps: 1. We first identified amino acid sequences of all intracellular proteins and 9-mers from a human protein database (UniprotKB). 2. Next, we use a computational program to predict binding affinities and half-lives of all possible combinations of 9-mers and 13 HLA-A and HLA-B alleles encompassing major HLA supertypes; combinations passing affinity and half-life criteria are classified as potential complexes. 3. For each potential pMHC complex, we identify potential similar complexes; two complexes are considered similar if they have the same HLA allele and 5 or more identical amino acids at identical positions. 4. We classify a complex as a potential target if combined mRNA expression of the complex, and similar complexes is higher in breast cancer tissues (TCGA data) compared to normal essential tissues (GTEx data) in at least 5% of the cancer samples. Results We extracted all intracellular protein sequences available in the UniprotKB, from which all 9-mer peptides were derived. Hundreds of thousands of potential pMHC complexes using binding predictions for 13 HLA A and B alleles were identified. For each potential MHC binding peptide, similar amino acid sequences were identified. Combined mRNA expression analysis resulted in the discovery of potential targets, which we further prioritized on the basis of likelihood of off-target effects and breast cancer expression levels of the peptide's gene(s) of origin. While we identified several thousand potential breast cancer specific targets, the number of targets with low likelihood of cross-reactivity is incredibly small. Conclusions We have developed a method to identify potential cancer-specific pMHC complexes and to predict potential off-targets of the drugs targeting them, which is essential to the development of safe drugs. Using this method, we examined the landscape of potential breast cancer specific targets derived from the human peptidome. Our examination showed that finding targets with low likelihood of cross-reactivity may be extremely difficult. Nonetheless, our method could help in prioritizing pMHC targets based on the evaluation of the potential cross-reactivity risk. Citation Format: Ankur Dhanik, Wen Zhang, Jessica Kirshner, Douglas MacDonald, Susan Croll, Calvin Lin, Gavin Thurston, Andrew Murphy. Exploration of the human peptidome for cancer-specific peptide-MHC targets. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 1088. doi:10.1158/1538-7445.AM2015-1088