Molecular structure of the ESCRT III-based archaeal CdvAB cell division machinery

Abstract

Most prokaryotes divide using filaments of the tubulin-like FtsZ protein, while some archaea employ instead ESCRT-III-like proteins and their filaments for cell division and cytokinesis. The alternative archaeal system comprises Cdv proteins and is thought to bear some resemblance to ESCRT-III-based membrane remodelling in other domains of life, including eukaryotes, especially during abscission. Here we present biochemical, crystallographic and cryo-EM studies of the Sulfolobus Cdv machinery. CdvA, an early non-ESCRT component, adopts a PRC-domain/coiled-coil fold and polymerises into long double-stranded helical filaments, mainly via hydrophobic interfaces. Monomeric CdvB adopts the canonical ESCRT-III fold in both a closed and a distinct "semi-open" conformation. Soluble CdvB2 filaments are composed of subunits in the closed state, appearing to transition to the open, active state only when polymerised on membranes. Short N-terminal amphipathic helices in all CdvB paralogues, B, B1 and B2, mediate membrane binding and are required for liposome recruitment in vitro. We provide a molecular overview of archaeal ESCRT-III-based cytokinesis machinery, the definitive demonstration that CdvB proteins are bona fide ESCRT-III homologues and we reveal the molecular basis for membrane engagement. Thus, we illuminate conserved principles of ESCRT-mediated membrane remodelling and extend them to an anciently diverged archaeal lineage. Significance statementMembrane remodelling by ESCRT-III proteins is a fundamental and conserved process across the tree of life. The archaeal ESCRT-III-based cell division system (Cdv) drives cytokinesis in many archaeal groups, yet the molecular architecture of its components remained unknown, making it difficult to decipher the molecular mechanisms employed for cell division and cytokinesis in these organisms. We present structures of the Cdv machinery in Sulfolobus organisms that have been used previously to study the cell division process. We show that CdvA forms unexpected antiparallel helical filaments, while the ESCRT-III homologues retain the canonical fold we show they share with eukaryotic proteins. We demonstrate that N-terminal helices mediate membrane binding and that membrane contact, rather than polymerisation alone, likely triggers activation of Cdv ESCRT-IIIs.