In the final decades of the 19th century, scientists showed in rapid succession that many of the worst diseases to afflict humanity were the work of bacteria—germs. Leprosy, gonorrhea, diphtheria, tuberculosis, plague, cholera, dysentery: Barely a year went by without assigning an infamous illness to a newly identified microbe. This concept, where one germ causes one disease, has influenced the way we think about infections ever since, and it implies an obvious solution: Remove the bug, and cure the sickness.

But the links between microbes and poor health can be more complicated. Our bodies are naturally home to tens of trillions of bacteria. Most are benign, or even beneficial. But often, these so-called microbiomes can shift into a negative state. For example, inflamed guts tend to house an unusually large number of bacteria from the Enterobacteriaceae family (pronounced En-ter-oh-back-tee-ree-ay-see-ay, and hereafter just “enteros”). There’s no villain in this scenario, no single antagonist as there would be in the case of tuberculosis or cholera. The enteros are part of a normal gut; it’s the same old community, just altered.

These kinds of shifts are harder to rectify. For a start, it’s often unclear if the enteros cause the inflammation, if the inflammation changes the microbes, or both. Even if the microbes are responsible, how do you fix that? Dietary changes are typically too imprecise. Antibiotics are too crude, killing off beneficial microbes while suppressing the problematic ones.

But Sebastian Winter, from the University of Texas Southwestern Medical Center, has an alternative. His team showed that the blooming enteros rely on enzymes that, in turn, depend on the metal molybdenum. A related metal—tungsten—can take the place of molybdenum, and stop those enzymes from working properly.

By feeding mice small amounts of tungsten salts, Winter’s team managed to specifically prevent the growth of enteros, while leaving other microbes unaffected. Best of all, the tungsten treatment spared the enteros under normal conditions, suppressing them only in the context of an inflamed gut. It’s a far more precise and subtle way of changing the microbiome than, say, blasting it with antibiotics. It involves gentle nudges rather than killing blows.

To be clear, no one knows if this would work in people. “We can cure inflammatory bowel disease in mice, and that’s the best we can say at this point,” Winter says. “We’re far away from having a treatment. And of course, tungsten is toxic, so this is not an endorsement that people with IBD should drink tungsten-contaminated water. But we can now screen for molecules that have the same activity without the toxicity.”

“It shows that the microbiome can indeed be edited if we understand how certain organisms thrive in a given environment,” says Manuela Raffatellu, from the University of California at San Diego. And that understanding, she adds, takes years of work.

Many teams, for example, have shown how enteros both bloom in inflamed guts, and trigger inflammation themselves. And Winter’s team has uncovered several of the tricks behind their ascension. These microbes are typically found in low numbers because they need oxygen to grow, and the gut is an oxygen-free world. But during inflammation, oxygen leaks through, and its presence allows enteros to devour a chemical called formate, produced by other gut microbes. “They can eat the scraps off the table,” Winter says. Inflammation also causes host cells to release nitrates, and the enteros can “breathe” using these instead of oxygen.

These discoveries all pointed to a single Achilles’ heel. It turned out that the enzymes that allow enteros to process both formate and nitrates—that allow them to eat and breathe—use a single atom of molybdenum. Tungsten is similar to molybdenum, sitting right beneath it in the periodic table. It’s chemically similar enough that it can substitute for the other metal in the bacterial enzymes, but different enough that once this happens, those enzymes are dead. They don’t work, and the enteros can’t grow.

That’s what Winter and his colleagues found. Team members Wenhan Zhu and Maria Winter fed tungsten salts to mice that had been previously dosed with DSS—a chemical that inflames the gut. Enteros would normally bloom vigorously in such conditions, but the tungsten reduced the numbers by almost a million times. It didn’t, however, affect the rest of the microbiome.

Cathryn Nagler, from the University of Chicago, says the results are intriguing, but she’s disappointed that the team used DSS. It’s often used to simulate inflammation “because it’s quick and easy,” she says, but it’s also crude, and doesn’t capture the full complexity of IBD. Winter acknowledges this, but he says that his team also proved the effectiveness of tungsten in rodents whose guts had been inflamed in other ways. They even showed that tungsten reduces inflammation in mice that had been loaded with the microbiomes of people with IBD. “That’s the closest we can get [to showing that this might work in people] without doing clinical trials,” he says.

“It’s a very important advance,” says Gary Wu, from the University of Pennsylvania. First, it shows that the altered bacterial communities that are associated with IBD are actually perpetuating the disease rather than just going along for the ride. Second, it hints at a way of changing those communities “in a way that is nonlethal to bacteria, unlike antibiotics.”

Other scientists are working on similar approaches. In 2015, I wrote about a team from the Cleveland Clinic who are searching for chemicals that reduce the risk of heart disease by targeting gut bacteria. Those microbes transform nutrients in our diet into chemicals that can slow the breakdown of cholesterol, causing fat to build up in our arteries. By shutting down the enzymes behind this process, it might be possible to spare our hearts—and again, without actually killing any microbes. This is what medicine might increasingly look like: less a war against specific germs, and more a series of gentle nudges applied to entire communities.