Novel fiber structure and protein VII overexpression identified in adenovirus D10

Researchers studying human adenovirus D10 have mapped its three-dimensional structure at unprecedented resolution and uncovered an unexpected pattern in how the virus makes its proteins—findings that could reshape how scientists develop this virus as a treatment for cancer and other diseases.

Adenoviruses are double-stranded DNA viruses that cause everything from common colds to eye infections. Scientists have long been interested in weaponizing them as therapeutic vectors—essentially repurposing the virus to deliver cancer-fighting payloads directly to tumor cells. The most widely used version in clinical work is adenovirus C5, but it has a major liability: most people have already been exposed to it, so their immune systems recognize and neutralize it before it can reach its target. Adenovirus D10, by contrast, is rare in human populations and has low affinity for the receptors that other adenoviruses use to enter cells, making it a potentially cleaner slate for engineering toward specific tissues.

To understand whether D10 could safely and effectively deliver a therapeutic payload, researchers used cryo-electron microscopy to freeze viral particles and image them at atomic resolution. They analyzed 5,524 individual virus particles and reconstructed the capsid—the protein shell that holds the viral DNA—at 3.3 angstroms of resolution. The overall architecture matched what scientists already knew about adenovirus structure: a roughly spherical shell made of hexon proteins forming the facets, penton proteins anchoring the corners, and fiber proteins projecting outward like antennae to grab onto host cells. But when the team focused on the fiber shaft—the thin stalk that extends from the penton base—they found something new. Embedded in the shaft at residues 111 through 161 was a previously uncharacterized structure they named the "umbrella motif." The motif consists of two beta-strands arranged so that one flares away from the main fiber structure, creating a small protrusion with a solvent-exposed loop. The apex of this umbrella carries a positive electrical charge from lysine and arginine residues, while the loop itself contains a negatively charged glutamate. This electrostatic landscape suggests the motif might act as a secondary binding site for host cell receptors, complementing the fiber's globular knob domain at its tip.

The structural findings alone would be noteworthy, but the transcriptomic analysis revealed something more puzzling. The researchers infected human esophageal cancer cells with D10 and sequenced the viral RNA at 24, 48, and 72 hours after infection. They found that the virus produced the expected early, intermediate, and late genes in the expected temporal sequence. But protein VII—a DNA-binding protein that helps compact the viral genome inside the capsid—showed an unusual pattern. The virus produced two distinct transcripts coding for protein VII: one for the precursor form, which is normal, and another for a mature form that typically only arises after the viral protease cleaves the precursor. At 72 hours post-infection, D10 was making precursor protein VII transcripts at a ratio of roughly 4.3 to 1 compared to mature protein VII transcripts. More striking, protein VII accounted for nearly 15 percent of all viral transcripts at late timepoints, far exceeding the percentage of transcripts for major structural proteins like hexon and fiber. This pattern differed markedly from adenovirus C5, where mature protein VII transcripts are vanishingly rare.

The researchers also found that D10 produces higher levels of transcripts for protein Mu, another DNA-binding protein, compared to C5. Both proteins work alongside protein V to condense the viral DNA into dense nucleoprotein complexes called adenosomes. The overproduction of these DNA-binding proteins in D10 raises a question the authors cannot yet answer: does D10 actually make more of these proteins than C5, or do the transcripts not translate into proportionally more protein? And if it does make more, why? The virus might require extra DNA-binding capacity for some aspect of its lifecycle that remains unknown, or the high transcript levels might reflect inefficient translation or rapid protein turnover.

The researchers also identified three transcripts that do not match any known adenovirus genes—hypothetical transcripts predicted by computational models but never before observed in actual viral infection. Two of these are heavily spliced, using canonical splice sites, which suggests they are genuine viral products rather than sequencing artifacts. Their function, if any, remains entirely unclear.

The authors emphasize that their work was conducted in esophageal cancer cells, not the eye cells that D10 naturally infects, so the transcriptomic profile might differ in the virus's native tissue. They call for proteomic studies to confirm whether the high transcript levels for protein VII and protein Mu translate into correspondingly high protein abundance, and they note that the hypothetical transcripts need validation to determine whether they actually encode functional proteins. The findings underscore a broader point: not all adenoviruses are interchangeable. Each serotype carries its own structural quirks and transcriptional signatures, and those differences could have consequences for safety and efficacy when the virus is engineered as a therapeutic. Before D10 or any other low-seroprevalence adenovirus moves into clinical trials as an oncolytic agent or vaccine vector, its biology must be thoroughly mapped.