Giant Viruses Caught Carrying Their Own Membrane-Remodelling Machinery
Viruses have long been cast as the ultimate freeloaders of the living world: stripped-down genetic packages that depend on host molecular machinery for nearly every step of their existence. Nowhere has that dependence seemed more absolute than in the way enveloped viruses exit their host cells. To pinch themselves free of a membrane, they almost universally commandeer the cell’s own membrane-fission device, the ESCRT machinery. A study now published in Nature Microbiology shows that this tidy division of labour has at least one spectacular exception. Medvedeva, Guyet, Koonin and their colleagues report the discovery of large eukaryotic DNA viruses that carry genes encoding core ESCRT components of their own. Homologues of ESCRT-III, the filament-forming module that constricts membrane necks, and of Vps4, the ATPase that dismantles and recycles those filaments, were identified in mirusviruses — a phylum of plankton-infecting viruses described only in 2023 — and in certain members of Nucleocytoviricota, the phylum that encompasses the giant viruses. These pathogens of single-celled eukaryotes did not merely borrow the cell’s membrane-snipping tools; at some point in the deep past they acquired the blueprints and have been carrying them ever since.
The ESCRT machinery — short for endosomal sorting complexes required for transport — is among the most versatile nanomachines in eukaryotic cells, and its architecture explains why. The system is assembled from a handful of multiprotein modules, ESCRT-0 through ESCRT-III, together with accessory factors. In the classic endosomal pathway of yeast and animals, ESCRT-0, -I and -II recognize ubiquitin-tagged cargo and concentrate it into patches on the limiting membrane of endosomes, poised to be swallowed inward as intraluminal vesicles. The decisive act belongs to ESCRT-III. Its subunits — the CHMP proteins of animals and fungi, with Snf7 as the yeast prototype — circulate in the cytoplasm in auto-inhibited folds and, on activation, expose amphipathic helices that tether them to the membrane surface. There they polymerize into spiralling filaments on the cytosolic face of the bud neck, drawing the membrane progressively tighter until the two bilayers within it merge and the connection is severed. The AAA+ ATPase Vps4 then completes the cycle: through MIT-domain interactions with the tail of each ESCRT-III subunit, its hexameric ring pulls the spent subunits through a central pore, releasing them back into the soluble pool for another round of assembly. Without Vps4, a locked ESCRT-III lattice stalls the entire machine.
The core pair of ESCRT-III and Vps4 is also ancient. Bacteria largely do without it, but many archaea use ESCRT-III, assisted by the protein CdvA, to divide their cells, while animals deploy the same duo for the final abscission step of cytokinesis, for sealing wounds in the plasma membrane, for rebuilding the nuclear envelope after mitosis and for shedding membrane from remodelled protrusions. Enveloped viruses, meanwhile, turned ESCRT into a cornerstone of their exit strategy. HIV carries a short “late domain” within its p6 protein that recruits Tsg101, an ESCRT-I subunit, and the adaptor ALIX, summoning the downstream budding apparatus to the viral assembly site. The matrix protein VP40 of Ebola viruses and budding factors of numerous other enveloped viruses exploit the same cellular machinery through analogous recruiting motifs. In every documented case, however, the virus contributed only a docking tag; the engines of constriction and scission themselves were invariably host-encoded. No virus had ever been shown to carry ESCRT components in its own genome, which is precisely what makes the new report a genuine exception rather than an incremental extension of an established pattern.
The exception emerged from a systematic survey of large eukaryotic DNA virus genomes. Screening the genomes of viruses in the phylum Nucleocytoviricota and of mirusviruses, Medvedeva, Guyet, Koonin and colleagues detected homologues of both ESCRT-III and Vps4 in a defined subset of these viruses — specifically, Nucleocytoviricota viruses known to infect unicellular eukaryotes, alongside the mirusviruses themselves. The pairing of the two genes is the salient feature of the finding. A lone ESCRT-III gene on a viral genome would be biologically awkward: filament polymerization without ATP-driven disassembly is a one-shot, potentially self-poisoning reaction. Together with Vps4, however, ESCRT-III forms the minimal, self-contained engine of membrane scission — the same two-part module that performs fission in archaeal division and in countless processes within eukaryotic cells. The presence of that complete module on viral genomes implies a functional unit under selection rather than a decaying passenger sequence, and it suggests that these viruses can, in principle, run membrane-remodelling cycles on their own schedule, with their own parts, inside the cells they infect.
Half of the story concerns the viruses in which the genes were found. Mirusviruses belong to Mirusviricota, a phylum announced in 2023 after metagenomic analyses of plankton collected across the global ocean revealed an abundant and previously hidden branch of large DNA viruses. The name derives from the Latin mirus, meaning strange or astonishing, and the group has lived up to it: mirusviruses combine herpesvirus-like genes for DNA replication and transcription with an expanded functional repertoire reminiscent of the giant Nucleocytoviricota, occupying an evolutionary position that bridges two great assemblages of eukaryotic DNA viruses. They parasitize unicellular eukaryotes in sunlit surface waters from pole to pole, infecting the plankton that anchor marine food webs and shape the biological carbon pump. The new analysis adds a further twist to their profile: the ESCRT genes of mirusviruses, the authors conclude, appear to have been hijacked from a host cell early in the evolution of the lineage, rather than acquired piecemeal by individual viruses in recent times.
That conclusion carries evolutionary weight. Large DNA viruses are notorious genomic mosaics, assembling their repertoires through horizontal gene transfer from hosts, from other viruses and from the organelles of the cells they infect. Retention of a captured gene, however, is not automatic: genes that confer no advantage are typically eroded and lost over evolutionary time. If mirusviruses acquired ESCRT-III and Vps4 early and have maintained them across their subsequent diversification, the acquisition must have paid for itself repeatedly. Plausible payoffs follow from what is known about infection biology. A virus equipped with its own scission machinery could bud from intracellular membranes without waiting for, or competing with, the host’s endogenous ESCRT supply. It could remodel internal membranes to construct replication compartments or wrapping sites tailored to its own assembly programme. In unicellular eukaryotes, whose ESCRT systems may be tuned to endosomal trafficking and cell division rather than viral egress, such independence could be decisive. For the Nucleocytoviricota members involved, separate acquisition events from their own hosts would indicate that the same evolutionary solution was found more than once.
What the viral ESCRT module actually does during infection remains to be demonstrated experimentally, but the physics of membrane remodelling narrows the possibilities. ESCRT-III filaments assembled on the cytosolic face of an endosomal, Golgi-derived or other cellular membrane could constrict a bud carrying viral genomes and envelope proteins, exactly as host ESCRT constricts HIV particles at the plasma membrane. The virus-encoded Vps4 would then reset the lattice, allowing the next particle to form and enabling the rapid, repeated budding that productive infection demands. Because the genes sit on viral chromosomes, their expression could be timed to the infection cycle and steered to particular subcellular neighbourhoods, and the proteins could evolve biochemical preferences for the unusual lipid and protein composition of virus-modified membranes that host enzymes would lack. Validation will require the standard arsenal of modern virology: purification and reconstitution of the viral proteins on model membranes, assays of polymerization, ATP hydrolysis and vesicle scission in vitro, transcriptomic and proteomic evidence of expression during infection, and ultimately direct visualization of budding intermediates in infected planktonic hosts.
The discovery also resonates beyond virology. Viruses are increasingly viewed not simply as thieves of cellular genes but as reservoirs and brokers of genetic novelty, participants in an ongoing exchange that has shaped the evolution of cells themselves. Finding the irreducible core of the cell’s membrane-fission engine inside viral genomes raises the converse question of directionality: if viruses can capture ESCRT components and make them their own, some cellular membrane-remodelling systems might, in part, bear the imprint of ancient viral genes that flowed back into host lineages. The result likewise sharpens the evolutionary portrait of the large DNA viruses, a group whose genomes have long blurred the boundary between viral and cellular gene space. And it underscores a recurring lesson of comparative genomics: in archaea and eukaryotes alike, the functional heart of membrane scission is the same ESCRT-III–Vps4 duo, stripped of much of the accessory complexity that surrounds it in animals. A virus that carries exactly that heart appears to have converged on the same minimal solution that cells themselves settled on.
For now, the viral ESCRT genes rest on comparative genomics, and the burden of proof has passed to the laboratory bench. Determining whether the mirusvirus and Nucleocytoviricota proteins constrict and sever membranes on their own, how tightly their expression is woven into the infection cycle, and what advantage they confer in the plankton ecosystems where these viruses roam will define the next phase of work. The stakes extend past a single curious gene pair. Membrane remodelling underlies cytokinesis, endosomal trafficking, wound repair and viral egress throughout biology, and each new carrier of the ESCRT engine clarifies how this ancient machine has been redistributed across the tree of life. It also recalibrates expectations for what viral genomes can contain. The maxim that enveloped viruses always borrow their budding machinery, taught for decades, now carries a caveat written in the genomes of viruses drifting through the ocean: a few of them, at some point, stopped borrowing and started building.
Subject of Research: Discovery of ESCRT-III and Vps4 homologues encoded in the genomes of mirusviruses and certain Nucleocytoviricota large eukaryotic DNA viruses that infect unicellular eukaryotes, indicating that some viruses carry their own membrane-remodelling machinery, with mirusviruses having hijacked host ESCRT machinery early in their evolution.
Subject of Research: Biology
Article Title: Large eukaryotic DNA viruses encode ESCRT machinery for membrane remodelling
Article References: Medvedeva, S., Guyet, U., Koonin, E. V., Delmont, T. O., & Krupovic, M. (2026). Large eukaryotic DNA viruses encode ESCRT machinery for membrane remodelling. Nature Microbiology. https://doi.org/10.1038/s41564-026-02470-9
Image Credits: AI Generated
DOI: 10.1038/s41564-026-02470-9
Keywords: ESCRT machinery, ESCRT-III, Vps4, mirusviruses, Nucleocytoviricota, giant viruses, membrane remodelling, membrane scission, viral evolution, horizontal gene transfer, unicellular eukaryotes, plankton viruses
Cite Scienmag News
APA
MLA
Chicago
Audrey B. (August 29, 2026). Large DNA viruses carry their own membrane remodelling machinery. Scienmag. https://scienmag.com/large-dna-viruses-carry-their-own-membrane-remodelling-machinery/
Audrey B. “Large DNA viruses carry their own membrane remodelling machinery.” Scienmag, 29 August 2026, https://scienmag.com/large-dna-viruses-carry-their-own-membrane-remodelling-machinery/. Accessed 29 August 2026.
Audrey B. “Large DNA viruses carry their own membrane remodelling machinery.” Scienmag. August 29, 2026. https://scienmag.com/large-dna-viruses-carry-their-own-membrane-remodelling-machinery/
Copy citation
Download RIS
Tags: ESCRT components in virusesGiant DNA viruseslarge eukaryotic DNA viruseslarge eukaryotic virusesmembrane remodelling in virus life cyclemembrane remodelling machinerymirusvirusesmirusviruses membrane genesNucleocytoviricotaNucleocytoviricota virus biologyviral dependence on host machineryviral evolution of cellular machineryviral evolution of membrane machineryviral gene acquisition and inheritanceviral membrane fission mechanismsviral membrane remodelling genesvirus gene encoding ESCRT proteinsvirus independence from host machineryvirus-host membrane interactions
