What Does the Shape Space of T Cell Epitopes Look Like?

Shape space is a decades-old conceptual model of antibody-antigen interactions that underlies antigenic maps used to trace viral evolution. Here, we apply this concept to T cell receptors (TCRs) and the peptide-MHC complexes (pMHCs) that they recognize. We start by reviewing the history of shape space and its deep connections to concepts in statistical physics (energy landscapes) and computer science (complexity theory). Leveraging these connections, we propose a model in which TCR-pMHC binding relies on multiple, possibly conflicting, structural constraints-implying that pMHCs recognized by the same TCR occupy several disjoint, non-convex regions in shape space. Our model makes two central predictions: (1) even small pMHC structure alterations can have major effects on immunogenicity; (2) two pMHCs recognized by the same TCR can have very different structures. We show that published TCR-pMHC interaction data generated by mutagenesis assays lend some support for this idea. We conclude by discussing implications of a multispecific and non-convex T cell epitope shape space for the prediction of immune responses in cancer immunotherapy and other applications.