4. Discussion

In this study, we identified a novel virus with circular ssDNA genome, HthCRESSV1, associated with the marine oomycete H. thermoambigua. Although viruses with linear ssDNA genomes classified in the recently created family Oomyviridae (phylum Cressdnaviricota, class Arfiviricetes, order Lineavirales) have been predicted to infect oomycetes based on the presence of related EVEs in the genomes of diverse oomycetes species (Sabanadzovic et al. 2025), to the best of our knowledge, HthCRESSV1 represents the first experimentally characterized cressdnaviricot infecting an oomycete. Indeed, all viruses infecting marine fungi or oomycetes isolated up to now have RNA genomes (Nerva et al. 2016, 2019, Botella et al. 2020, Botella and Jung 2021). This is in line with the fact that most known mycoviruses, regardless of the ecosystem they inhabit, have dsRNA or ssRNA genomes, whereas ssDNA mycoviruses are relatively rare (Kondo et al. 2022). Notably, HthCRESSV1 is unrelated to viruses of the family Oomyviridae (Sabanadzovic et al. 2025), which have linear genomes, encode a typical jelly-roll capsid protein, and based on the Rep phylogeny are placed within the class Arfiviricetes. By contrast, Rep phylogeny firmly places HthCRESSV1-like viruses within the class Repensiviricetes, as a separate family, which we propose naming ‘Mitodnaviridae’.

Arguably the most unexpected feature of HthCRESSV1 is its association with mitochondria. Detection of HthCRESSV1-like EVEs in oomycete species from different genera suggests that the association with mitochondria is lasting, rather than transient. The only other previously known group of viruses associated with mitochondria are ssRNA viruses of the family Mitoviridae, which replicate in mitochondria of filamentous fungi (Polashock and Hillman 1994, Polashock et al. 1997). Phylogenetic analysis of the Rep positions HthCRESSV1 at the base of the clade including related EVEs, suggesting that the latter derive from the integration HthCRESSV1-like viruses into mitochondrial genomes. Notably, Southern blot analysis indicated that the host strain does not carry an integrated copy of HthCRESSV1, suggesting that integration into the mitochondrial genome is not an obligatory step of virus replication. HthCRESSV1 and related EVEs are most closely related to unclassified and uncultivated viruses, some of which encode typical jelly-roll capsid proteins, suggesting that HthCRESSV1 lineage has evolved from bona fide cressdnaviricots following the replacement of the ancestral capsid protein gene with the one encoding the membrane-associated ORF2 protein, an α-helical membrane protein with five N-terminal transmembrane domains and a C-terminal coiled-coil region. Structural modelling suggests that this protein can form multimeric complexes within membranes, likely as heptamers. Homologues of ORF2 were not detected in known members of the phylum Cressdnaviricota but were identified in several oomycete mitochondrial genomes (Martin et al. 2007, Derevnina et al. 2015, Winkworth et al. 2020, 2022, Morgan and Tartar 2023) suggesting an evolutionary relationship with mobile mitochondrial elements or integrated viral fragments. These homologues often occur without adjacent rep genes, implying independent acquisition through horizontal transfer or secondary loss of rep. The tight association of the HthCRESSV1 genome with mitochondrial DNA, coupled with the absence of nuclear integration, supports the hypothesis that replication occurs exclusively within mitochondria. This is further reinforced by TEM analysis, which revealed vesicle-like structures potentially associated with viral replication.

Functionally, ORF2 likely contributes to membrane dynamics. If the viral genome is encapsulated within membrane-bound vesicles, ORF2 may mediate vesicle formation, budding, or membrane penetration events that facilitate virus spread, replication, genome packaging, or evasion of mitochondrial antiviral defences (Miller and Krijnse-Locker 2008, Hsu et al. 2010). Alternatively, if the virus remains confined within the mitochondrial compartment, like in the case of the RNA mitoviruses, ORF2 may remodel inner membrane structures such as cristae or mitovesicles to create specialized niches for replication (Kopek et al. 2007, den Boon and Ahlquist 2010). The presence of a C-terminal coiled-coil domain supports a role in oligomerization or scaffolding, common features of membrane-associated structural or assembly proteins (Burkhard et al. 2001, Lupas and Bassler 2017). Given its distinctive combination of features and phylogenetically isolated position, ORF2 likely represents a novel membrane-active viral factor, possibly acquired through horizontal transfer from a mitochondrial mobile element or host organellar DNA.

Overall, the evolutionary history of HthCRESSV1 is consistent with the current views on the cressdnaviricot evolution, which involves recurrent recombination between plasmid-like replicons leading to repeated acquisition, exchange, and loss of capsid protein genes (Koonin et al. 2015, Kazlauskas et al. 2018, Desingu and Nagarajan 2022, Krupovic and Koonin 2026). In this framework, the Rep phylogeny firmly roots HthCRESSV1 within Repensiviricetes, attesting to its viral origin. By contrast, replacement of the canonical jelly-roll capsid protein gene with HP in the ancestor of the proposed family ‘Mitodnaviridae’ could be associated with the relocation to mitochondria as the replicative compartment. In a sense, this transition might be considered as return to the roots, considering that cressdnaviricots are believed to have evolved from bacterial plasmids (Kazlauskas et al. 2019), whereas mitochondria represent domesticated alphaproteobacterial endosymbionts.

From a virological perspective, most DNA viruses, with the notable exception of certain large dsDNA viruses from the phylum Nucleocytoviricota, replicate in the nucleus. It remains unclear whether minimal cressdnaviricot viruses, such as HthCRESSV1, possess the capacity to actively cross the mitochondrial double membrane. Information on viruses that directly interact with mitochondria is extremely limited. For example, mitovirus-derived sequences have been identified in both the nuclear and mitochondrial genomes of several vascular plants (Hong et al. 1998, Xu et al. 2014, Bruenn et al. 2015, Nibert et al. 2018). Among infectious DNA viruses, a reverse-transcribing hepatitis B virus (HBV) not only integrates into the host nuclear genome but has also been shown to deliver HBV RNA into mitochondria of hepatocellular carcinoma cells, where it may participate in mtDNA integration (Giosa et al. 2023). This process is mediated by polynucleotide phosphorylase (PNPase), which facilitates RNA transport into mitochondria. Given the RNA specificity of PNPase, its involvement in the translocation of the DNA genome of HthCRESSV1 into mitochondria is unlikely. Further studies using immunoelectron microscopy or other imaging approaches will be required to determine whether HthCRESSV1 localizes to the mitochondrial matrix or the intermembrane space.

Consistent with this proposed mitochondrial association, indirect evidence supporting the interaction between HthCRESSV1 and mitochondria includes its highly efficient vertical transmission via zoospores and variation in host growth across temperatures. The species epithet of H. thermoambigua reflects its ambiguous optimal growth temperature (Maia et al. 2022); comparative growth assays showed that, compared to the virus-free strains, the virus-infected strain BD651 exhibited a reduced growth overall while maintaining relatively similar growth at higher temperatures. Mitochondrial genomes have been implicated in thermal adaptation in several organisms, including P. infestans, Drosophila melanogaster, and Saccharomyces cerevisiae (Camus et al. 2017, Li et al. 2019, Shen et al. 2022). Although HthCRESSV1 is not integrated into mtDNA, its association with mitochondria may influence organelle function and contribute to reduced growth rates and elevated optimal growth temperature. In this context, mitochondrial RNA mitoviruses provide a useful comparison. Although evolutionarily unrelated to HthCRESSV1 and typically associated with cryptic infections, they share a common intracellular niche and are transmitted vertically through spores very efficiently (Polashock et al. 1997, Čermáková et al. 2017, Schoebel et al. 2017). In some cases, mitoviruses have been linked to altered host phenotypes; for example, infection by Sclerotinia sclerotiorum mitovirus 1 (SsMV1) in S. sclerotiorum is associated with reduced growth and hypovirulence, alongside mitochondrial abnormalities (Xu et al. 2014).

Our attempts to eliminate HthCRESSV1 using ribavirin were unsuccessful, and the virus remained stably transmitted through zoospores. Similar persistence has been reported for other cressdnaviricots infecting fungi (Wang et al. 2024). Notably, successful re-infection has been demonstrated for certain viruses such as SsHADV-1 using purified virions or infectious clones (Yu et al. 2010, Wang et al. 2024). Developing a re-infection system for HthCRESSV1, potentially through mitochondrial transfection approaches, will be essential to clarify its biological effects on the host.