Epitope-based cancer immunotherapies have been intensely investigated for decades, but have thus far met with mixed results in the clinical setting. Recent years have, however, witnessed a resurgence of interest based on advances in high-throughput sequencing approaches that allow rapid identification of cancer- associated, patient-specific mutations having the potential of being recognized as epitopes (neo-epitopes). Previous studies defined binding affinity associated with immunogenicity for HLA class I MHC restricted CTL responses of viral and non-self origin. However, it is not clear if the same threshold is applicable to identify epitopes in self-antigens, such as the ones recognized in autoimmunity and cancer. Additional factors expected to play a crucial role in the immunogenicity of cancer neo-epitopes include the propensity of epitopes to be derived by natural processing, and subsequent capacity of being recognized by TCR. While several online tools are available for predicting processing and TCR recognition, the relative role of each of these factors (HLA binding, processing, TCR recognition) has not been analyzed in the cancer setting. Likewise, how the different online tools can be optimally combined to predict neo-epitopes has not been thoroughly addressed. Accordingly, our study will experimentally establish the threshold for HLA binding affinity of neo- epitopes, validate the suitability of HLA binding prediction methods in a cancer setting, and establish whether predicted binding affinities can be suitably used in lieu of experimentally measured binding constants. We will also evaluate the predictive value of additional variables globally affecting neo-epitope immunogenicity, including allele-specific HLA binding thresholds, HLA-peptide binding stability, binding capacity of the somatic sequence from which a neo-epitope is derived, the predictive capacity of TAP and proteosomal cleavage algorithms and, to account for neo-epitope TCR interface effects, we will evaluate two alternative schemes based on analysis of the residue type occupying putative TCR contact positions. Finally, we will explore combining all neo-epitope immunogenicity predictors into a single pipeline and provide experimental validation of its predictive efficacy. The proposed research will provide 1) the first in-depth characterization of the relative importance of different variables affecting neo-epitope immunogenicity, 2) the first integration of predictive tools for each of these variables into a single pipeline, and 3) freely available online access of a newly developed pipeline to the scientific community. This is highly innovative, since no such resource is currently available for epitope immunogenicity in general, and for cancer neo-epitopes in particular.
The proposed research is intended to provide in-depth characterization of the importance of several variables affecting cancer neo-epitope immunogenicity. These findings in turn will allow the development of bioinformatic tools, and a unified analysis pipeline, towards the identification of cancer neo-epitopes. Identification of such epitopes is expected to allow quick identification of cancer-associated, patient-specific, mutations potentially recognized as epitopes, with direct applicability to the development of efficacious therapeutics.