Conformational Dynamics Linking Kinase and Endonuclease Domains Explain the Divergent Function of IRE1 Paralogues

Abstract

The ER stress sensors IRE1α and IRE1β have dual kinase and endonuclease activities that mediate cellular proteostasis by splicing Xbp1 mRNA and degrading other ER‐targeted transcripts. Although the two paralogues share a high degree of sequence homology, the epithelial cell‐specific paralogue IRE1β, which is expressed at mucosal surfaces, has impaired kinase activity and phosphorylation that restrict stress‐induced activation of the endonuclease domain. To uncover structural features that explain these differences, we used normal mode analysis to build dynamic correlation networks describing active and inactive conformations of IRE1 kinase‐endonuclease domains. For human IRE1α, we identified a fingerprint of intra‐ and interdomain dynamic couplings that distinguish between the two conformations. Mutations that disrupted IRE1α endonuclease activity perturbed the dynamic networks of the active conformation and enhanced features of an inactive conformation, thus validating our approach. Strikingly, the dynamic networks of human IRE1β when modeled in an active conformation resembled the inactive IRE1α network. A set of non‐conserved amino acids were found to distinguish between the dynamic networks linking kinase‐endonuclease domains for the two IRE1 paralogues. In most cases, however, these divergent network‐determining amino acids were found only in IRE1β sequences from mammals. In lower vertebrates, the key amino acids and the dynamic couplings of IRE1α active state were conserved in IRE1β. Thus, IRE1β evolved distinct conformational dynamics and enzymatic activity for a role in mucus barrier function that is unique to higher vertebrates.