💻 bioinformatics

A hierarchical orthology framework reveals viral carbohydrate-active genes across the global virosphere

This study introduces VirGenes, a hierarchical orthology framework that integrates sequence, structural, and functional data to uncover a vast, previously underappreciated diversity of viral carbohydrate-active enzymes across the global virosphere, revealing their complex evolutionary origins and significant roles in virus-host interactions.

Original authors: Meng, L., Zhang, R., De Castro, C., Uchiyama, I., Kanehisa, M., Ogata, H.

Published 2026-08-07

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Meng, L., Zhang, R., De Castro, C., Uchiyama, I., Kanehisa, M., Ogata, H.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine the microscopic world as a bustling, chaotic city where tiny invaders called viruses are constantly trying to break into the homes of bacteria and other cells. To succeed, these invaders often need to wear a disguise or carry a special tool to unlock the front door. For a long time, scientists thought viruses were like master thieves who only borrowed tools from their victims, relying entirely on the host's machinery to build their disguises. However, we now know that some viruses are actually master craftsmen in their own right, carrying their own blueprints to build complex sugar-based structures. These sugar-manipulating tools are called carbohydrate-active enzymes, or CAZymes. Think of them as the virus's personal set of lockpicks, paintbrushes, and glue guns that help them stick to cells, hide from the immune system, or even break down the cell walls to escape. The big mystery has always been: how many of these tools do viruses actually carry, and where did they get them? Are they just borrowing from their hosts, or have they built their own unique toolkit over millions of years?

Enter a team of researchers who decided to build a massive, organized library to solve this puzzle. They created a new database called VirGenes, which acts like a giant filing system for viral genes. Instead of just looking at the genetic code (the letters) to find matches, which is like trying to find a book by its title alone, they also looked at the 3D shape of the proteins (the story inside the book). This is crucial because viruses evolve so fast that their genetic "titles" often change beyond recognition, but their "stories" (their shapes) often stay the same. By using this clever, multi-layered approach, the researchers scanned the entire known "virosphere" (the world of viruses) and discovered something amazing: viruses are packed with sugar-manipulating tools they didn't know about.

They found 558 distinct groups of these viral genes, which belong to 102 different families of CAZymes. That's a lot of tools! Some viruses, particularly those that infect bacteria (bacteriophages), are absolute hoarders of these genes. Two specific families of bacteria-eating viruses, Kleczkowskaviridae and Pootjesviridae, were found to carry more than 10 of these sugar-tools in a single virus genome. Even the giant viruses that infect eukaryotic cells (like amoebas) were found to have a rich collection, with the Mimiviridae family showing up as a major player.

But the most exciting part of the story is how these viruses got their tools. The researchers traced the family trees of these genes and found that viruses are not just copying their hosts; they are constantly swapping genes in a high-stakes game of evolutionary tag. They found strong evidence that viruses frequently steal genes directly from bacteria to create enzymes that break down bacterial cell walls (like the GH23 and GH24 families). It's like a virus stealing a lockpick from a house to break into a different house, then modifying it to work even better. In fact, for some of these enzyme families, more than 80% of the members found in the database were viral, suggesting that viruses have become the primary inventors of these specific tools.

The researchers also discovered that viruses are hiding some of their best tools in plain sight. They found 34 groups of viral genes that looked completely different from any known sugar-tool when you just read their genetic code, yet when you looked at their 3D shapes, they were perfect matches for known enzymes. One standout example involved giant viruses that carry a specific shape called a "five-bladed β-propeller," which looks exactly like a sugar-digesting machine (specifically the GH130 family), even though the genetic code gave no hint of it. When the team tested these shapes with computer simulations, they found that these viral proteins could likely grab onto sugar molecules just as well as their known cellular cousins.

In short, this paper suggests that viruses are far more independent and creative than we thought. They aren't just passive borrowers; they are active participants in the global sugar economy, constantly acquiring, modifying, and even inventing new ways to manipulate carbohydrates. By building this new library, VirGenes, the researchers have opened the door to understanding how these tiny invaders survive, evolve, and interact with the world around them, revealing a hidden layer of complexity in the ongoing evolutionary arms race between viruses and their hosts.

Problem
The functional annotation of viral proteins remains a significant bottleneck in virology, with less than 30% of viral protein families currently assigned functional roles. This limitation is particularly acute for carbohydrate-active enzymes (CAZymes), which shape virus–host interactions by modifying virion structures, host surfaces, and extracellular glycans. While recent discoveries suggest that some viruses, particularly giant viruses, encode their own CAZymes, the diversity and evolutionary origins of these viral carbohydrate-active systems remain poorly understood. Existing bioinformatic frameworks rely heavily on sequence similarity, which often fails to detect remote homologs due to the rapid evolution and high variability of viral proteins. Consequently, many viral genes with conserved structural folds but low sequence identity to known CAZymes are missed, leaving a "dark matter" of viral functional potential unexplored.

Methodology
To address these gaps, the authors developed VirGenes, a hierarchical database of viral orthologous groups constructed from the KEGG viral gene dataset. The framework integrates three levels of homology detection:

Viral Sequence Clusters (VSCs): Generated via sequence similarity networks using MMseqs2 and the Markov Clustering (MCL) algorithm. A critical preprocessing step involved identifying and segmenting putative polyprotein precursors (e.g., coronavirus ORF1ab) into mature peptide regions to prevent heterogeneous clustering.
Viral Remote Clusters (VRCs): Formed by grouping VSCs based on remote homology detection using HHblits and HHsearch, capturing relationships beyond direct sequence similarity.
Viral Fold Clusters (VFCs): Created by mapping VSCs to protein structures (using AlphaFold predictions and PDB data) and performing all-against-all structural alignments with Foldseek.

The authors applied this framework to screen for viral CAZymes (including glycosyltransferases, glycoside hydrolases, polysaccharide lyases, carbohydrate esterases, auxiliary activity enzymes, and carbohydrate-binding modules). They employed a multi-step detection pipeline involving BLAST and HMMER searches against the CAZy and dbCAN databases, followed by phylogenetic analysis to trace evolutionary origins (specifically horizontal gene transfer events) and structural analysis to identify CAZyme-like folds lacking sequence homology.

Key Contributions

VirGenes Database: A publicly accessible resource (https://www.genome.jp/vogdb/) that provides a unified, hierarchical framework connecting viral sequence space, remote homology, and structural fold space. It integrates KEGG Orthology (KO) assignments and metadata from Virus-Host DB and ICTV.
Systematic CAZyme Discovery: The identification of 558 viral orthologous groups (VSCs) containing 8,807 CAZyme-encoding genes spanning 102 CAZyme families. This represents a significant expansion over previous detection methods.
Structural Annotation of "Dark Matter": The demonstration that structural similarity can reveal CAZyme-like functions in viral genes that lack detectable sequence similarity. Specifically, the study identified numerous hypothetical viral sequences (particularly from giant viruses) that possess glycoside hydrolase-like five-bladed $\beta$ -propeller folds.

Results

Distribution: Viral CAZyme repertoires are widespread, with particularly enriched sets in dsDNA viral lineages. Two bacteriophage families, Kleczkowskaviridae and Pootjesviridae, encode more than 10 CAZymes per genome, while the giant virus family Mimiviridae also shows high densities of CAZyme-encoding genes.
Evolutionary Origins: Phylogenetic analyses reveal divergent evolutionary histories. There is evidence of frequent horizontal gene transfer (HGT) from bacteria to viruses, particularly for endolysin genes (e.g., GH23, GH24), suggesting a viral strategy in the evolutionary arms race with cellular hosts. Conversely, some families (e.g., AA10) show diversification within phages after a single ancestral acquisition, while others show sporadic transfers from eukaryotes.
Structural Insights: In the structural space of VirGenes, approximately 85% of viral genes with CAZyme-like folds lacked detectable sequence similarity to annotated CAZymes. A notable finding is the presence of $\beta$ -propeller folds in giant viruses that structurally mimic GH130 (glycoside hydrolase) and PL22 families. Molecular docking of a representative viral $\beta$ -propeller (VSC_11007) suggested a preference for $\beta$ -1,3-mannobiose, consistent with GH130 substrate recognition, despite the protein being annotated as hypothetical.

Significance
The paper posits that viral carbohydrate-active systems exemplify how "distributed innovations," constrained by ancient structural folds, collectively build the functional complexity of the global virosphere. By integrating sequence, structural, and functional evidence, VirGenes moves beyond the limitations of sequence-only searches to illuminate the functional potential of viral "dark matter." The study suggests that viral CAZyme repertoires are not sporadic but represent adaptive strategies shaped by selective pressures, such as host cell wall degradation and glycan synthesis. The framework establishes a scalable foundation for interpreting viral gene function, bridging the gap between protein classification and evolutionary/ecological context, and providing a resource for future studies on virus–host interactions.

More like this