Structures of distantly related interacting protein homologs are less divergent than non‐interacting homologs

Homologous proteins can display high structural variation due to evolutionary divergence at low sequence identity. This classical inverse relationship between sequence identity and structural similarity, established many years ago, has remained true between homologous proteins of known structure over time. However, a large number of heteromeric proteins also exist in the structural data bank, where the interacting subunits belong to the same fold and maintain low sequence identity between themselves. It is not clear if there is any selection pressure to deviate from the inverse sequence–structure relationship for such interacting distant homologs, in comparison to pairs of homologs which are not known to interact. We examined 12,824 fold pairs of interacting homologs of known structure, which includes both heteromers and multi‐domain proteins. These were compared with monomeric proteins, resulting in 26,082 fold pairs as a dataset of non‐interacting homologous systems. Interacting homologs were found to retain higher structural similarity than non‐interacting homologs at diminishing sequence identity in a statistically significant manner. Interacting homologs are more similar in their 3D structures than non‐interacting homologs and have a preference towards symmetric association. There appears to be a structural constraint between remote homologs due to this commitment.