Assessing structure-function impacts on Vitellogenin by leveraging allelic variant occurring in honey bee subspecies Apis mellifera meliffera

Computational advances involving artificial intelligence (AI) and successful experimental state-of-the-art structure determination can provide detailed pictures of large and complex protein structures, and their variations. A standout case is Vitellogenin (Vg) derived from the honey bee ( Apis mellifera ). Vg is an essential protein for reproduction in almost all egg-laying animals, and can in addition regulate behavior and provide immunological support in some species, including the honey bee. Information is limited in terms of the structure-function relationships that underlie Vg’s pleiotropic functions, at least in part because this protein is not expressed in the best developed gene-editing models, such as fruit flies and mice. However, naturally occurring allelic variation in Vg can provide some insight, i.e., into which changes (mutations) are allowable (present at some frequency) vs. likely not allowable (not present or only at low frequency). Here, we leverage a unique dataset of 1,086 fully sequenced Vg alleles from honey bees in 15 countries. We identify a population-specific 9 nucleotide deletion in a locally endangered honey bee subspecies ( A. m. mellifera ) that impacts a loop structure in a central Vg domain. Due to the A. m. mellifera population history of near extinction and human intervention, an assessment of this Vg variant is not only theoretically interesting but also relevant for subspecies conservation efforts. Using structural bioinformatics, molecular dynamics simulations, and a transformer-based indel predictor (IndeLLM), we demonstrate that Vg protein structure and stability can be maintained despite the deletion. Our approach also reveals the dynamic nature of specific regions in Vg for the first time. Generalizable results may extend to other egg-laying animals of ecological and economic importance.