Compositional bias and mimicry toward the nonself proteome in immunodominant T cell epitopes of self and nonself antigens.

We investigated whether and how molecular mimicry affects the shaping of the helper T cell repertoire. We implemented an algorithm that measures the probability of mimicry between epitopes of known immunogenicity and self or nonself proteomes. This algorithm yields 'similarity profiles', which represent the probability of matching between all contiguous overlapping peptides of the antigen under examination and those in the proteome(s) considered. Similarity profiles between helper T cell epitopes (of self or microbial antigens and allergens) and human or microbial SWISSPROT collections were produced. For each antigen, both collections yielded largely overlapping profiles, demonstrating that self-nonself discrimination does not rely on qualitative features that distinguish human from microbial peptides. However, epitopes whose probability of mimicry with self or nonself prevails are, respectively, tolerated or immunodominant and coexist within the same (auto-)antigen regardless of its self/nonself nature. Epitopes (on self and nonself antigens) can cross-stimulate T cells at increasing potency as their similarity with nonself augments. Mimicry, rather than complicating self-nonself discrimination, assists in the shaping of the immune repertoire and helps define the defensive or autoreactive potential of a T cell. Being a predictor of epitope immunogenicity, it bears relevance to vaccine design.