Influenza A Dismantles a Small Nuclear Compartment, Protein-by-Protein Map of Infected Cells Shows

Influenza A virus borrows surface receptors to slip into human cells, then dismantles a small, obscure nuclear compartment called paraspeckles from the inside out, according to a new protein-by-protein map of intact, infected cells [published Monday in Nature Microbiology](https://www.nature.com/articles/s41564-026-02416-1?error=cookies_not_supported&code=f192d256-791a-4771-9e3d-649e03455a2b).

Paraspeckles are tiny, membraneless droplets inside the nucleus that help cells manage stress and control which genes get read. They were first described in the last decade and remain poorly understood. The new map shows influenza A virus progressively taking them apart across multiple cell lines during infection.

The work, from a team at EMBL Hamburg and the Leibniz Research Institute for Molecular Pharmacology (FMP) in Berlin, is the first time researchers have tracked the physical handshakes between viral and human proteins directly inside intact influenza A-infected cells at a scale detailed enough to model how the contact points fit together structurally.

"Before this, we had a list of suspects," co-senior author Jan Kosinski said in the EMBL release. "Now we can see who is actually touching whom in a living, infected cell."

The viral nucleoprotein (NP) and a smaller viral protein called NS1 both bind to paraspeckle proteins, anchoring the virus to the compartment. A viral enzyme called PA-X, an endonuclease the virus uses to chew up host RNA, then goes after the long non-coding RNA that holds the paraspeckle together. The compartment falls apart as the RNA scaffolding is removed. Paraspeckles normally buffer cells against stress, so stripping them likely makes the host more vulnerable to the rest of the viral assault.

The team combined in-cell cross-linking mass spectrometry, a technique that freezes proteins in place with chemical "staples" while the cell is still alive, with AlphaFold-based structural modeling and functional assays. AlphaFold's structure predictions let the team move from raw crosslinks to a model of each contact. Most prior maps of influenza A protein–protein interactions relied on breaking cells open first, which destroyed the natural geography of the nucleus and other compartments. The new pipeline keeps that geography intact.

The team also flagged a second pattern: a set of host factors involved in maturing the sugar decorations, or glycoforms, on the viral surface protein haemagglutinin (HA) as it moves through the cell's membrane-bound ER-Golgi system. These are potential vulnerabilities for drugs that try to slow the virus down before it leaves the cell.

Whether the paraspeckle disassembly is a direct viral target or a downstream symptom of the broader transcriptional chaos influenza A causes is not yet known. If the latter, the hijacking frame softens into a stress response, and the next round of work will need to test which paraspeckle proteins the virus is grabbing on purpose. Kosinski and colleagues flag the same question in the EMBL release and in the GEN coverage.

Seasonal influenza kills up to 650,000 people globally each year and sends three to five million more to hospital, according to the EMBL release. Influenza A's replication depends on interactions between as many as 14 viral proteins and hundreds of human factors, and most prior maps could only catch those interactions after the cell had already been broken apart. The in-cell workflow changes that for any virus–host pair, not just flu.

Kosinski's team has used the same method on other pathogens in earlier work and is now extending it. The researchers told GEN the workflow is reusable for the next dangerous strain, including H5N1, and for any virus–host pairing where the compartment context is the point. A follow-up paper testing which of the new contacts survive in cells infected with a lab-adapted H5N1 strain is the team's stated next step.