Exploring the double-stranded DNA viral landscape in eukaryotic genomes

This study introduces a computational framework that screened over 37,000 eukaryotic genomes to identify hundreds of thousands of endogenous double-stranded DNA viral regions, revealing their widespread presence across diverse species and uncovering numerous novel virus-host associations that expand our understanding of the endogenous dsDNA virosphere.

The Gist

Imagine your genome (your body's instruction manual) as a massive, ancient library. For decades, scientists knew that this library contained some "stolen" pages from viruses, specifically a type called retroviruses (like HIV's cousins). These were well-documented, like famous bookmarks everyone knew about.

But what about the rest of the library? What about the massive, double-stranded DNA viruses (the "heavy hitters" of the virus world, like the ones that cause herpes or giant amoeba-eating viruses)? Scientists suspected these had also left their mark, but they were like ghosts in the library—hard to find and poorly understood.

This paper is like a team of high-tech librarians who built a super-powered metal detector to scan the entire library of 37,000 different eukaryotic species (animals, plants, fungi, and single-celled organisms) to find these hidden viral "ghosts."

Here is what they found, explained simply:

1. The "Metal Detector" Scan

The researchers created a computer program that acts like a metal detector. It knows exactly what the "metal" (viral DNA) sounds like compared to the "wood" (normal animal or plant DNA).

• The Result: They scanned 37,254 different genomes and found 781,111 viral fragments.
• The Surprise: These weren't just in a few species; they were in 19% of all the genomes they checked. That's like finding hidden treasure in nearly one out of every five houses you walk into.

2. The "Viral Real Estate"

Some of these genomes are absolutely drowning in viral DNA.

• The Bivalve Boom: In some freshwater mussels, viral DNA makes up 16% of their entire genome. Imagine if 16% of your body's instructions were actually written by a virus.
• The Protist Party: Single-celled organisms like Trichomonas (a parasite) also had huge chunks of viral DNA (over 12%).
• The Vertebrate Chill: Humans and other vertebrates (like fish and birds) had much less (less than 1%). It seems our advanced immune systems are like strict security guards that kicked the viral squatters out long ago, whereas invertebrates (like insects and mussels) are more like open houses where viruses have been moving in for millions of years.

3. The "Squatters" vs. The "Guests"

The team figured out two ways these viruses ended up in the DNA:

• The "Integration" (The Squatter): The virus got stuck in the DNA and stayed there, becoming part of the host's family tree. This is what happened in 98% of the cases. It's like a virus crashing a party, getting drunk, and then moving into the guest room permanently.
• The "Contig" (The Guest): Sometimes, the virus was just sitting on a piece of DNA that looked like a whole virus, but hadn't necessarily merged with the host yet. This was rare (only 2%).

4. New "Ghost Stories" (New Discoveries)

This is the most exciting part. Before this study, we only knew of about 43 specific pairs of "Virus X infects Animal Y."

• The Expansion: This study found 144 new potential pairs where a virus might infect an animal, but we've never actually caught the virus in the wild to prove it.
• The Example: They found evidence that a specific type of virus (Lavidaviridae), which was previously thought to only infect single-celled protists, has actually been living inside the DNA of insects (specifically a type of fly) and sponges. It's like finding out that a shark species thought to live only in the ocean has actually been living in a freshwater river for thousands of years.

5. The "Library Renovation"

The study also found that these viral fragments aren't just junk.

• The Tools: Many of these viral fragments contain "tools" (genes) that the host might have borrowed. For example, some insects seem to have kept viral genes that help them fight off other viruses or manage their cell structures.
• The Evolution: It's as if the library didn't just get stolen pages; it got useful pages that the librarians (the host) decided to keep and use to improve the building.

The Big Picture

Think of this study as a massive archaeological dig. For a long time, we only knew about the "famous ruins" (retroviruses). This paper used a new map and new tools to uncover a whole hidden city of viral ruins buried inside the DNA of almost every type of life on Earth.

It tells us that viruses and their hosts have been in a constant, messy, and intimate dance for billions of years. Sometimes the virus wins and takes over the host's genome; sometimes the host steals the virus's best ideas to survive. We are all, in a way, a little bit of a virus, and this paper helps us read the footnotes of that story.

Technical Summary

1. Problem Statement

While endogenous viral elements (EVEs) derived from retroviruses are well-documented and known to constitute significant portions of eukaryotic genomes (e.g., ~8% of the human genome), the landscape of endogenous double-stranded DNA (dsDNA) viruses remains largely underexplored. Although specific cases of dsDNA virus integration (e.g., HHV-6, mavirus) and domestication (e.g., Bracoviruses in wasps) are known, there is no comprehensive, systematic framework to quantify the extent of dsDNA viral integration across the broad diversity of eukaryotic life. Furthermore, the "endo-virosphere" (the collection of endogenous viral elements) is poorly characterized compared to the free-living virosphere, limiting our understanding of virus-host coevolution and the evolutionary impact of dsDNA viruses on host genomes.

2. Methodology

The authors developed a computational framework to identify dsDNA viral regions (VRs) within eukaryotic genome assemblies. The pipeline involved the following key steps:

• Dataset: Analysis of 37,254 eukaryotic genome assemblies (as of Feb 2024) covering Metazoa, Fungi, Viridiplantae, and Protists.
• Reference Construction:
  • Curated 2,126 reference dsDNA viral genomes representing 11 major viral taxa (e.g., Adenoviridae, Herpesvirales, Megaviricetes, Polinton-like viruses).
  • Constructed Hidden Markov Models (HMMs) for 7,755 orthologous groups (OGs).
  • Defined Viral Orthologous Groups (VOGs): OGs with no significant similarity to cellular references.
  • Defined Taxonomic Orthologous Groups (TOGs): VOGs specific to a single viral major taxon.
  • Constructed a separate set of cellular reference HMMs from curated cellular databases (KEGG, Pfam) to filter out host contamination.
• Detection Pipeline:
  1. ORF Prediction: Predicted Open Reading Frames in eukaryotic contigs.
  2. Screening: Scanned ORFs against VOGs and TOGs. Identified "virus-contig pairs" where contigs contained viral signatures.
  3. VR Delineation: Defined VRs as regions enriched with clustered viral ORFs. A region was classified as a VR if it contained multiple distinct viral hits from a single taxon, with limited intervening cellular ORFs (max 10 ORFs, max 10 kb distance).
  4. Classification:
     • Integration-like VRs: Regions where viral sequences appear integrated into the host genome (shorter than the contig, mixed with cellular ORFs).
     • Viral contig-like VRs: Regions where the viral sequence occupies >50% of the contig and contains <20% cellular ORFs (suggesting unintegrated viral DNA or assembly artifacts).
  5. High-Confidence VRs (HCVRs): A stringent subset filtered for assembly quality (BUSCO scores), taxonomic consistency (dominant taxon signals), and presence of conserved OGs.

3. Key Results

A. Scale and Distribution

• Total Discovery: Identified 781,111 VRs across 7,103 assemblies (19% of all analyzed genomes).
• Taxonomic Bias:
  • Metazoans: Dominated the findings (96.98% of VRs), particularly in invertebrates.
  • Protists: 2.58% of VRs, but with high diversity.
  • Fungi/Plants: Very low prevalence (<0.5% combined), likely due to cell wall barriers or stronger antiviral defenses.
• Genomic Impact:
  • VRs occupy >10% of genomes in diverse invertebrates (e.g., 16.2% in the bivalve Unio delphinus, 14.7% in the insect Neoaliturus tenellus).
  • In contrast, vertebrates show low VR content (average 0.014%), with humans at ~0.004%.
  • No correlation was found between VR abundance and assembly completeness, suggesting findings are not artifacts of low-quality data.

B. Biological Characteristics

• Coding Density: VRs exhibit significantly higher ORF density (20–40% higher) than surrounding host genomic regions, confirming their viral origin.
• Sequence Composition: Tetranucleotide frequency (TNF) analysis showed that VRs from Herpesvirales and Lavidaviridae often retain distinct viral signatures, while those from Orthopolintovirales and Pokkesviricetes have compositions more similar to their hosts (suggesting long-term integration and amelioration).
• Degradation: VRs generally contain shorter ORFs than reference viruses, indicating widespread pseudogenization.
• Functionality: HCVRs are enriched in replication genes (e.g., DNA polymerase, capsid proteins) but also contain host-interaction factors like GOLGB1 (Golgi architecture) and BIRC family genes (apoptosis inhibition), suggesting potential roles in host-virus coevolution.

C. Novel Associations and Lineages

• Expanded Associations: The study recovered 29 of 43 previously known virus-eukaryote associations (based on isolation) and identified 144 new candidate associations lacking isolation evidence.
• Specific Discoveries:
  • Lavidaviridae in Insects: Identified integration of Lavidaviridae (virophages) in the pomace fly Lordiphosa mommai and sponge Ephydatia muelleri, challenging the dogma that virophages only infect protists.
  • Mirusviricota in Metazoans: Found Mirusviricota signals in a parrot genome, expanding the host range beyond unicellular eukaryotes.
  • Deep Viral Clades: Phylogenetic analysis revealed "VR-only" clades (branches containing only endogenous sequences) within Alloherpesviridae, Mriyaviricetes, and others, suggesting the existence of entirely novel viral lineages not yet captured by metagenomics or isolation.

4. Significance and Implications

• Baseline for the Endo-Virosphere: This study provides the first large-scale quantitative baseline for dsDNA viral elements in eukaryotes, revealing that they are far more pervasive in invertebrates and protists than previously thought.
• Evolutionary Insight: The high abundance of VRs in invertebrates (which lack adaptive immunity) versus their scarcity in vertebrates suggests that host immune systems play a critical role in purging endogenous viral elements.
• Discovery Engine: The catalog of VRs serves as a "gateway" to hidden viral diversity. The identification of VR-only clades implies that many viral lineages may exist primarily or exclusively as endogenous elements, evading traditional culture-based or metagenomic detection.
• Functional Potential: The presence of host-interaction genes within VRs highlights the potential for viral domestication, where viral genes are co-opted by hosts for essential functions (e.g., immunity, development).
• Resource: The authors provide a comprehensive catalog of VRs, HMMs, and phylogenetic data, offering a foundation for future studies on virus-host coevolution and the functional significance of endogenous viral elements.

In summary, this paper fundamentally shifts the perspective on dsDNA viruses, demonstrating that they are not just transient pathogens but major, persistent architects of eukaryotic genome evolution, particularly in non-vertebrate lineages.