This has not been a good week for glyphosate, the active ingredient in Roundup and other herbicides. On Friday, the World Health Organization (WHO) announced that it had classified glyphosate, the United States’ most widely-used pesticide, as “probably carcinogenic to humans.”

Now, the chemical has another strike against it. A study published today by the American Society of Microbiology’s journal mBio has linked glyphosate and two other widely-used herbicides–2,4-D and dicamba–to one of the most pressing public health crises of our time: antibiotic resistance.

This study found that exposure to these herbicides in their commercial forms changed the way bacteria responded to a number of antibiotics, including ampicillin, ciprofloxacin, and tetracycline–drugs widely used to treat a range of deadly diseases.

Dicamba, 2,4-D, and glyphosate have been in use for decades, so why have their antibacterial-resistance effects not been documented before? As the study’s lead author, Jack Heinemann, professor of genetics at the University of Canterbury in New Zealand, explains, when pesticides are tested for adverse effects, “it’s the lethal toxicity that people focus on.” In other words, how much of the chemical will kill an organism.

“What makes our study different, is that it is looking at a sub-lethal effect,” says Heinemann. “The effect we see requires that the bacteria stay alive.”

Previous studies done by other researchers have found that substances chemically similar to dicamba and 2,4-D can cause antibiotic resistance, Heinemann explains. So he and his colleagues decided to investigate whether these herbicides would produce similar effects. They added glyphosate to the study because it is chemically unlike the other two. But, to their surprise, it also produced some antibiotic resistance.

Heinemann explains that because these herbicides are not “supertoxic” to the bacteria the study tested–E. coli and Salmonella–they are not killed outright at levels typically used to kill weeds. Instead, the bacteria stay alive while activating proteins known as efflux pumps in order to rid themselves of toxins. And this defense mechanism can make the bacteria develop resistance to the threat from which it is defending itself.

Scientists know that overuse of antibiotics in humans can decrease their effectiveness. In the same way, says Heinemann, “exposure to these pesticides make the pathogens stronger.”

Although this study only looked at two laboratory strains of human pathogens, the antibiotics examined represent what he calls “broad classes” of drugs we’ve come to depend on to fight infections and the herbicides are three of the most-used worldwide.

Heinemann also notes that the different pesticides produced a variety of responses. While all three produced an antibacterial-resistant response to some of the antibiotics, some of the combinations his team tested produced no response and some increased the antibiotic’s effect.

Although the study is likely to be seen as controversial by some, University of Massachusetts Dartmouth assistant professor of biology, Dr. Mark Silby says it “followed established protocols” and the existing scientific literature supports its findings.

“This is a very carefully-designed study,” says Dr. Michael Hansen, a senior staff scientist at Consumers Union. “It’s incredibly important work showing the complexity of an effect that hadn’t been thought about before.” The mechanisms by which the bacteria respond to toxics–in this case herbicides–are already well-known, Hansen explains. What’s new and important is looking at non-lethal levels of exposure in combination with the antibiotics.

The weed-killers used in the study were purchased at a local store and were used at levels specified in use directions, which means the scientists were testing chemicals actually in use worldwide rather than a special laboratory sample of the active compound.

How could any of this affect people?

“These herbicides are now used at such a scale that we can almost use the term ubiquitous,” says Heinemann. For one, glyphosate is used on about 94 percent of the soybeans and 89 percent of the corn grown in the U.S, while 2,4-D is the third-most widely used herbicide in the U.S., while dicamba ranks fifth in use worldwide.

The levels at which the researchers saw effects were higher than the residues allowed on food, but below what is often used in rural settings, says Heinemann.

The results of Heinemann’s study suggest there is probably a small chance that exposure through food would produce these effects, but they could be a concern in areas where the pesticides are being applied, says Hansen. Thus, the people most likely to be affected are farmers, farmworkers, and other people who live in agricultural communities.

Also to consider is the approval earlier this year of a new pesticide that combines glyphosate and 2,4-D and soybean and cotton seeds genetically engineered to resist dicamba, all of which are expected to increase use of these pesticides.

Pesticide-induced antibiotic resistance could also affect honeybees since many commercial hives are now being treated with antibiotics. It’s possible, Heinemann says, that “comingling of antibiotics and herbicides could be compromising the effectiveness of those antibiotics,” and thus honeybee health.

Meanwhile, Monsanto says it disagrees with WHO’s announcement on glyphosate. “All labeled uses of glyphosate are safe for human health and supported by one of the most extensive worldwide human health databases ever compiled on an agricultural product,” the company says in a statement on its website.

Neither Monsanto nor other pesticide manufacturers have had the opportunity to respond to the new mBio study. But the Council for Biotechnology Information said on its website “GMO Answers” last month, that glyphosate had once been considered for use as an antibiotic but that “levels needed to kill microbes are relatively high, and resistance can develop readily.” In other words, the phenomenon Heinemann and colleagues observed is not entirely unexpected.

“A jigsaw puzzle is a good metaphor,” for how these effects fit together, says the scientist.

The next steps in this research will be to test additional bacteria and pure samples of the pesticides. But for now, it’s clear that “further work is needed,” says Hansen. “This is something we need to look at as we expand the use of these herbicides.”