Researchers at the European Molecular Biology Laboratory's Hamburg site, working with the Leibniz Research Institute for Molecular Pharmacology and Charité Berlin, have produced what they describe as the most detailed map yet of exactly how influenza A rewires an infected human cell — and it revealed a hijacking strategy nobody had documented before.

A technique built to catch the virus in the act

Most methods for studying virus-host interactions require breaking cells apart first, which creates a real problem: molecules that were never actually in contact during infection can artificially end up mixed together, while fragile, real interactions can fall apart before they're ever measured. To get around that, researchers Boris Bogdanow and Fan Liu adapted a technique called cross-linking mass spectrometry, which locks protein contacts in place at the exact moment they occur inside living, infected cells. The team then combined that data with a modified version of AlphaFold, the AI system that predicts protein structures, to reconstruct which viral and human proteins were actually touching each other during a real infection. The results were published July 20 in Nature Microbiology.

The discovery nobody was looking for

The map confirmed one known hijacking strategy — how the virus's surface protein, hemagglutinin, moves through a cell's internal transport system to get properly built. The second finding was the surprise. Inside the nucleus, the virus was found to systematically dissolve paraspeckles: small, droplet-like structures made of RNA and protein that normally serve as storage compartments for RNA-binding proteins involved in the cell's stress response. "What surprised us most was the paraspeckles," said Iuliia Kotova, the study's first author, who conducted the work as a predoctoral fellow in the Kosinski Group at EMBL Hamburg and is now at ETH Zurich.

"Watching these tiny organelles in the nucleus dissolve, consistently across every cell line and every flu strain we tested, told us this isn't a side effect of infection — it might be a strategy."— Iuliia Kotova, first author, EMBL Hamburg / ETH Zurich

Why dissolving them helps the virus twice over

When the paraspeckles broke apart, they released the RNA-binding proteins that had been stored inside — proteins the virus appears to co-opt to support its own replication. Group leader Jan Kosinski pointed to a possible second benefit: paraspeckles are also believed to play a role in the cell's antiviral gene regulation and stress responses, meaning dismantling them may simultaneously weaken the cell's own defenses while supplying the virus with raw material. Getting both benefits from a single action would make the strategy an efficient one for the virus to have evolved.

Why a new mechanism matters for drug design

Nearly all current antiviral flu drugs work by directly blocking the virus's replication enzymes — a strategy that, like antibiotic resistance in bacteria, gives the virus an evolutionary incentive to mutate around the drug over time. A treatment designed instead to protect paraspeckles from being dissolved would work through a completely different mechanism, targeting a host-cell vulnerability rather than the virus's own machinery directly — a route that may prove harder for the virus to evolve resistance against, though that remains to be tested. Researchers also note the approach could help in studying more dangerous strains, including H5N1, since the paraspeckle-dissolving behavior held consistently across every strain tested in this study.

What's still ahead

This is laboratory-stage research, not a treatment. Turning the discovery into an actual drug candidate would require confirming which specific viral proteins drive the paraspeckle disassembly, testing whether protecting those structures measurably reduces illness severity in animal models, and — years down the line — human safety trials. For now, the finding hands flu researchers a genuinely new mechanism to target, drawn from watching the virus operate inside intact, living cells rather than inferring its behavior after the fact.