Cross-linking inside living cells: a new lens on how influenza commandeers us
Every year, influenza A infects 3 to 5 million people worldwide and kills up to 650,000. For decades, virologists have studied how the virus hijacks human cells using the standard toolkit of molecular biology: break the cells open, extract the proteins, run the assays. The problem with that approach is that it destroys the very architecture the virus exploits.
A team led by researchers at EMBL Hamburg, the Leibniz Forschungsinstitut fur Molekulare Pharmakologie (FMP) Berlin, and Charite Berlin has now published the first large-scale map of direct protein contacts between influenza A and human proteins captured inside intact, infected cells. The work, published in Nature Microbiology (DOI: 10.1038/s41564-026-02416-1), reveals viral strategies that traditional methods had missed and it opens a methodological door for studying pandemic-potential strains like H5N1 with unprecedented resolution.
The problem with broken cells
When a virus infects a cell, it does not simply dump its proteins into a homogeneous soup. Viral proteins travel through specific compartments, the endoplasmic reticulum, the nucleus, the Golgi apparatus, and interact with host proteins in precise spatial and temporal sequences. Breaking the cell open collapses these compartments, turning a structured city into a pile of rubble. Proteins that were never in the same room can form spurious contacts; genuine but transient interactions vanish.
The team addressed this using a technique called in-cell cross-linking mass spectrometry (XL-MS), developed at FMP Berlin and adapted specifically for virus-infected cells. The method works by adding a chemical cross-linker that permeates living cells and covalently bonds any two proteins that are within a few angstroms of each other. The cross-linked pairs are then identified by mass spectrometry, yielding a snapshot of who was touching whom at the moment of fixation, all within the intact cellular environment.
“XL-MS allows us to capture protein-protein interactions directly in infected intact cells, while also providing structural information,” said Boris Bogdanow of Charite Berlin.
The team then fed these experimental cross-linking constraints into a modified version of AlphaFold, the protein structure prediction AI, to model how the viral and host proteins fit together at their contact sites.
Two viral strategies exposed
The map revealed two specific hijacking mechanisms that had been invisible in traditional assays.
The first involves hemagglutinin, the spike protein that influenza uses to enter cells. The researchers tracked hemagglutinin through the host cell’s internal processing network and found that the virus commandeers several human proteins, some with previously unknown functions, to fold and chemically modify the spike protein, ensuring it becomes functional. Without this host-assisted folding, the virus cannot produce infectious particles.
The second finding is more dramatic. The team observed that paraspeckles, small, droplet-like compartments in the cell nucleus involved in stress responses and antiviral gene regulation, dissolve during influenza A infection. This releases a pool of RNA-binding proteins that the virus may hijack to support its own replication.
“Watching these tiny organelles in the nucleus dissolve, consistently across every cell line and every flu strain we tested, told us this is not a side effect, it might be a strategy,” said Iuliia Kotova, the first author, now at ETH Zurich.
The dissolution of paraspeckles may serve a double purpose for the virus: stealing useful proteins for replication while simultaneously weakening the cell’s ability to mount an antiviral defense.
A method with pandemic reach
Although the study used a lab-adapted influenza strain, the methodology is the real breakthrough. It can be applied to any virus at any stage of infection, including strains with pandemic potential such as H5N1 avian influenza, which has been circulating widely in mammals and raising concerns about human adaptation.
“This study lays the groundwork to apply the methodology to viruses of potential pandemic relevance, such as H5N1,” Bogdanow said.
The approach also solves a long-standing problem in structural virology: many of the most important host-pathogen interactions are transient, weak, or compartment-specific, making them invisible to traditional methods. In-cell XL-MS captures them precisely because it does not require the proteins to be stable enough to survive cell lysis.
The map is now available as a resource for the field. It represents not just a catalog of interactions but a demonstration that the way we study viral infection, inside the intact cell, fundamentally changes what we can see.
Reference: Kotova et al., “Mapping in-cell protein contact sites reveals hijacking of paraspeckles during influenza A virus infection,” Nature Microbiology (2026). DOI: 10.1038/s41564-026-02416-1.