New study shows how bacteria adapted a virus-derived injection system to recognize and attach to many different types of cells. By systematically identifying thousands of rapidly evolving receptor-binding proteins, the researchers explain how these systems can be retargeted again and again in nature by swapping the part that binds to cells. The work not only solves a long-standing mystery about how these bacterial machines function, but also demonstrates that they can be engineered to deliver proteins into specific human cells, pointing to future biomedical and biotechnological applications.

Viruses attack nearly every living organism on Earth. To do so, they rely on highly specialized proteins that recognize and bind to receptors on the surface of target cells, a molecular arms race that drives constant evolution.

Now, a new study published in Nature Communications, led by Prof. Asaf Levy of the Faculty of Agriculture, Food and Environment at the Hebrew University of Jerusalem, reveals just how far this evolutionary creativity can go.

Working over several years, doctoral researcher Nimrod Nachmias, together with collaborators Zhiren Wang and Xiao Feng from Prof. Peng Jiang’s laboratory at the NHC Key Laboratory of Systems Biology of Pathogens in Beijing, uncovered a vast and previously hidden repertoire of receptor-binding proteins used by bacteria, many of them borrowed from viruses, plants, fungi, and even animals.

A viral legacy repurposed by bacteria
At the center of the discovery are extracellular contractile injection systems (eCISs), sophisticated, virus-like molecular machines derived from bacteriophage tails. While viruses use these structures to infect cells, many bacteria have repurposed them as toxin delivery devices, deploying them against competing host organisms such as insects, and likely against competing microbes.

“eCISs look like viral weapons, but they are now fully integrated into bacterial life,” the researchers explain. “They are used in ecological battles we are only beginning to understand. It's amazing to observe how a virus (bacteriophage) injecting its DNA into specific bacteria evolved into a bacterial tools injecting protein toxins into large diversity of host cells".

Cracking a long-standing mystery
For years, researchers have suspected that eCISs rely on specialized receptor-binding proteins, equivalent to viral spike proteins, to recognize their targets. But identifying these proteins proved extraordinarily difficult.
Like the coronavirus spike protein, these receptor-binding protein domains, which are called tail fiber proteins, evolve extremely rapidly, constantly changing shape to keep up with their targets in what is defined as a molecular arms race. Traditional search methods repeatedly failed to detect them.

To overcome this, the team developed a new computational algorithm capable of identifying these elusive genes across thousands of genomes.

The result was striking.

The researchers identified 3,445 eCIS tail fiber proteins encoded within 2,585 eCIS gene operons across 1,069 bacterial and archaeal species, the most comprehensive catalogue of its kind.

Evolution on overdrive
The study revealed that eCIS tail fibers are built from two distinct parts:

• a conserved “anchor” domain that attaches the fiber to the eCIS particle, and
• a highly variable receptor-binding domain that determines which cell types can be targeted.

Using structure prediction tools, the team classified these proteins 1,177 distinct domain fold families, many of which are predicted to bind sugars and proteins on the surface of bacterial or eukaryotic cells.

Remarkably, genetic evidence suggests that many of these domains were acquired through horizontal gene transfer — not only from other bacteria and viruses, but also from plants, fungi, and components of animal immune systems. Horizontal gene transfer is a common phenomenon in bacteria. However, such a frequent gene acquisition from so many diverse eukaryotes, into one specific gene (the tail fiber gene of different microbes), is rare in nature.

“This is accelerated evolution on steroids,” the researchers say. “Bacteria are essentially sampling the biological world for useful binding tools and repurposing them.”

From discovery to demonstration
To test whether these findings had real-world implications, the researchers selected a candidate tail fiber from a Paenibacillus eCIS that resembles hemagglutinin, a receptor-binding protein best known from influenza and measles viruses. Namely, they speculated that it can bind to human cells.

They engineered a chimeric eCIS particle, equipping it with this newly identified fiber, and showed that it could bind to and inject proteins into human THP-1 monocyte-like cells, while sparing other cell types.

Further experiments suggested that D-mannose, a sugar found on human cell surfaces, may act as a key receptor, partially blocking binding when added externally.

Electron microscopy images captured virus-like particles attaching to human cells moments before delivering their molecular payload.

A growing biotechnological frontier

While the team emphasizes that other groups, including several startups, are also exploring engineered eCIS systems, the scale of this discovery dramatically expands the toolkit available for future applications. This work uncovers thousands of naturally evolved receptor-binding proteins, and these could eventually be harnessed to deliver drugs, enzymes, or other therapeutic molecules into specific cell types.

A map for future exploration
Beyond its technological promise, the study opens fundamental biological questions.

“We still know very little about what many of these systems do in nature,” the researchers add. “Which cells do they target? Under what conditions are they deployed? This catalogue gives us, for the first time, a map to start answering those questions.”

By revealing the extraordinary diversity of receptor-binding domains hidden within bacterial genomes, the Hebrew University–led research highlights how ancient viral machinery continues to shape life and how nature’s solutions may inspire the next generation of biomedical tools.