AI-driven structural informatics reveals diverse glycosylation machinery across double-stranded DNA viruses
By integrating AI-driven structural informatics with genomic analyses, this study uncovers a vast, previously overlooked diversity of viral glycosyltransferases across double-stranded DNA viruses, including a novel jumbo-phage locus encoding a complex machinery for sulfated pentasaccharide DNA modification.
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 This is an AI-generated explanation of a preprint that has not been peer-reviewed.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
Viruses are often imagined as simple genetic parasites, tiny invaders that hijack a host cell's machinery to copy themselves and then burst out. For decades, scientists believed that when viruses decorated their own surfaces with sugar chains—a process known as glycosylation—they were entirely dependent on the host cell to do the work. In this view, the virus simply borrowed the host's tools, resulting in a viral coat that looked much like the cell it had infected. However, a growing body of evidence suggests that some large viruses are far more self-sufficient than previously thought. They carry their own blueprints for building complex sugars, hinting at a hidden, autonomous world of viral chemistry that has remained largely invisible because the genetic instructions for these tools have changed so much over time that standard computer searches can no longer recognize them.
A team of researchers at the University of Georgia has now peeled back this layer of invisibility using a new approach that looks at the shape of proteins rather than just their genetic code. By combining artificial intelligence with advanced structural mapping, they have uncovered a vast, previously unknown diversity of sugar-making machinery encoded in the DNA of double-stranded DNA viruses. Their work reveals that these viruses do not just rely on a few known methods; instead, they possess a rich, varied toolkit for modifying their own genetic material and their outer shells, including some mechanisms that have never been seen before.
The researchers began with a fundamental problem: traditional methods for identifying enzymes rely on comparing genetic sequences, but viral genes evolve so rapidly that their sequences often become unrecognizable, even when the proteins they produce still perform the same job. To bypass this limitation, the team turned to protein structure. Because the three-dimensional shape of a protein is often more stable than its genetic code, two proteins can look completely different in their DNA instructions yet fold into the same functional shape. The team used an AI-driven system to scan nearly 300,000 predicted viral protein structures, searching for the specific architectural folds associated with glycosyltransferases, the enzymes responsible for attaching sugar molecules.
This structural search recovered over 1,700 candidate proteins that had been missed by standard genetic analysis. When the researchers grouped these candidates by their shapes, they found a dominant cluster containing nearly 600 proteins, along with many smaller, distinct groups. Crucially, they tested their method by feeding in six known viral sugar-making enzymes; the system correctly grouped all of them into the main cluster, proving that the structural approach could successfully identify these tools even when their genetic signatures were weak. This success allowed the team to explore the "dark matter" of viral biology, finding that many of these newly discovered enzymes did not fit into any known family of sugar-making tools, suggesting they represent entirely new branches of the evolutionary tree.
One of the most striking discoveries came from a specific group of these enzymes, which the researchers found in a variety of viruses ranging from giant viruses that infect amoebas to large bacteriophages that infect bacteria. In some cases, these enzymes were located next to genes involved in building the virus's outer shell, suggesting they decorate the virus particle itself. In other cases, they were found alongside genes that modify the virus's own DNA. This led to a particularly fascinating finding in a massive bacteriophage known as a jumbo phage. The researchers identified a specific genetic region in this virus that contains five sugar-making enzymes working in a row, alongside enzymes that prepare the DNA for modification and others that add sulfate groups.
This arrangement suggests the virus has the capacity to build a complex, five-sugar chain and attach it directly to its DNA, then add a sulfate tag to the end. While scientists have long known that some viruses attach simple sugars to their DNA to hide from bacterial immune systems, the idea of a virus building a long, sulfated sugar chain on its own genetic material is a new concept. The structural analysis of the enzymes involved indicates that one enzyme likely attaches the first sugar to a modified DNA base, while the others extend the chain, creating a unique chemical signature that has never been demonstrated before.
Beyond this specific example, the researchers mapped the locations of these sugar-making genes across hundreds of different viral genomes. They found that the potential to build sugars is not limited to a few special cases but is spread widely across the viral world, appearing in diverse groups from ancient bacterial viruses to giant eukaryotic viruses. By comparing the genetic neighborhoods of these viral enzymes to known sugar-building pathways in bacteria and other organisms, they found that viral systems occupy distinct regions of the chemical landscape. Some viral clusters resemble the machinery used to build protective capsules around bacteria, while others look like systems for making complex surface coatings.
The study does not claim to have solved the exact chemical recipe for every virus found, nor does it prove that every predicted pathway is active in nature. Instead, it provides a structural map that points to where these activities are likely happening. The researchers suggest that these diverse viral systems may serve different purposes: some might protect the viral DNA from being destroyed by host defenses, while others could alter how the virus is recognized by the immune system. By revealing that viruses possess such a wide array of autonomous sugar-making tools, the work challenges the old view of viruses as simple genetic parasites and positions them as complex biological entities capable of sophisticated chemical engineering. This new perspective opens the door to discovering entirely new types of biological molecules and understanding how viruses have evolved to survive in a world full of defenses.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.