Gene behavior distinguishes viral from bacterial infections

New approach could gauge response to flu vaccine

Flu virus and MRSA bacteria

CASING A CULPRIT  Certain gene behavior changes in people can reveal whether an infection is caused by a virus, such as the H1N1 influenza virus (left), or bacteria, such as methicillin-resistant Staphylococcus aureus, or MRSA (right). 

NIAID/Flickr (CC BY 2.0)

Coughs, fevers and green mucus can accompany an infection, but most of the time, doctors can only guess whether the culprit is bacterial or viral. A new study points out a way to identify the perp.

An infection changes the behavior of the afflicted person’s genes, and that host response differs depending on whether bacteria or a virus is doing the damage, scientists report in the Dec. 15 Immunity. This virus-bacteria distinction could ultimately help doctors quickly figure out what ails a person, and whether antibiotics, antiviral drugs or just chicken soup and sleep is the best treatment.

To find the viral fingerprints, computational immunologist Purvesh Khatri of Stanford University and colleagues combed through a wide variety of publicly available datasets that included information about how human genes behaved after an infection of influenza, human rhinovirus and respiratory syncytial virus, or RSV. The researchers churned these diverse datasets through a series of sophisticated mathematical analyses, a process that ultimately pinpointed a consistent viral calling card — a list of nearly 400 genes, each of which grew either more or less active during a viral attack. Many of those genes make proteins known to be involved in virus responses and inflammation.

Khatri and colleagues then tested whether the behavior of these genes could distinguish viral infections from bacterial infections, or no infection at all. Looking at separate datasets that weren’t used to generate the gene list, the researchers found that the virus signature could predict whether a person was infected with a virus, ruling out a bacterial cause. “This was a very robust host response,” Khatri says.

That response is “the most solid signature people have found,” says systems immunologist Shai Shen-Orr of Technion-Israel Institute of Technology in Haifa. And because the signature was consistent across a wide array of studies, the results have added heft. “It’s like this has just been lying here, waiting for someone to pick up,” he says. “The signal just jumped out.”

Further studies could pinpoint whether the virus was flu, Khatri says. The behavior of just 11 genes served as a signature of flu, revealing whether a person was infected with influenza as opposed to other viruses. 

Using data from studies in which healthy people were infected with the flu, the researchers showed that the viral signature showed up hours before symptoms appeared. The method even caught an asymptomatic carrier — someone who didn’t feel ill but was shedding virus nonetheless. “We could pick up a person walking around, not showing any symptoms,” Khatri says.  

The analysis also offers clues about how well the flu vaccine worked. The body’s response to vaccines was similar to the response to an actual flu infection: The behavior of the 11 genes reliably changed in both cases, a result that could allow doctors to track the effectiveness of the flu shot.  

Khatri and his team have already turned up a curious difference between men and women. In men, the influenza signature peaked in response to a flu vaccine the first day after the shot. Women’s responses, however, peaked three days later. Some vaccine studies have waited several days to look for an immune response, and concluded that men have weaker reactions, Khatri says. But a simple difference in timing may explain those observations.

The strength of the study comes from its reliance on diverse and numerous datasets, says systems immunologist John Tsang of the National Institute of Allergy and Infectious Diseases in Bethesda, Md. “The key here is that after you look across all of them [the datasets], you see a coherent signal,” he says. “That’s the encouraging sign.”

Tsang cautions that before this method could be useful in clinics, it will need to be tested in a large prospective trial, one designed to capture changes in gene behavior in people over time.

Khatri and colleagues are already working to design a test that could tell doctors whether a patient has a viral or bacterial infection. Such a test would cut down on unnecessary antibiotic use, Khatri says. 

More Stories from Science News on Health & Medicine

From the Nature Index

Paid Content