The idea of using bacteria-infecting viruses called bacteriophages to kill specific microbes implicated in disease has been around for more than a century. But a major barrier to using phage therapies is that the microbiome is constantly evolving and using molecular strategies such as CRISPR to evade attack. Research published today (August 4) in Cell purports to have taken a step toward overcoming that hurdle by using a cocktail of phages that use multiple lines of attack against Klebsiella pneumonia bacteria, preventing them from evolving resistance to the mixture. The researchers behind the study report that their blend of phages successfully treated symptoms of inflammatory bowel diseases (IBDs) such as ulcerative colitis and Crohn’s disease in mice, and was well-tolerated in very early testing on healthy human volunteers.

“For the first time, we [were] able to develop a precision therapy that would target a group of commensals within this huge and divergent microbiome without impacting the surrounding critically important microbial ecosystem,” says study coauthor Eran Elinav, a microbiome researcher at the Weizmann Institute of Science in Israel and the German Cancer Research Center.

Elinav and his colleagues first performed genomic sequencing to characterize the bacteria in the microbiomes of 537 people with IBDs in four countries: France, Germany, Israel, and the US. The analysis revealed that Klebsiella pneumonia was one of the most common bacteria in the cohort, present in nearly 40 percent of the people across all four regions. Furthermore, this species of bacteria was more abundant in patients experiencing IBD flareups compared to those in remission, suggesting it plays an important role in disease.

While Elinav says that other bacteria such as E. coli may also contribute to IBDs, K. pneumonia’s  pervasiveness throughout the cohort led the team to focus on it in their subsequent analyses. They further characterized the strains of K. pneumonia represented in stool samples from the people with IBDs, and through sequencing and computational analyses, identified a particular family, which they call the Kp2 clade, that was strongly associated with IBDs.

The researchers then evaluated how powerfully some of the representative strains within the Kp2 clade induced inflammatory responses in the guts of germ-free mice, looking at hallmarks of IBD such as high levels of certain cytokines. This enabled them to identify so-called driver strains of bacteria—those that most significantly contribute to disease—while differentiating them from less important strains, the authors write in the study.

The next step was to identify phages that target the driver strains, which Elinav describes as a painstaking fishing expedition that took five years. Starting with thousands of phages from sources such as dental waste and gut microbiomes, the researchers used an iterative process to isolate phages that attack K. pneumonia. When the bacteria inevitably evolved to escape those, the team identified phages that would attack the resistant mutants. Continuing back and forth in this manner, they wound up with 41 phages that attacked K. pneumonia with a range of strategies.

See “How Commensal Gut Bacteria Keep Pathogens in Check

The researchers then tested various combinations of the phages in mice infected with the most inflammation-inducing Kp2 strains and found a combination of five phages that best prevented bacterial growth. The final mixture represented phages with a range of attack strategies that prevented the bacteria from developing resistance to all of them. Finally, they investigated the ability of the five-phage mixture to suppress the disease-associated inflammatory response.

Paul Bollyky, an infectious disease specialist at Stanford University who was not involved in the research, says that he can envision other researchers being eager to adopt the approach of identifying bacteria associated with disease and then using phages to remove them. “The idea that you could actually engineer or modify the gut microbiome in particular, selective ways almost like you might do surgery in other tissues in your body is really novel and remarkable,” he says. 

As a step in that direction, Elinav’s team needed to check that the phages were stable and safe for humans. The researchers began by simulating the acidic environment in the stomach to see if the phages would tolerate it, but they didn’t. “This ex vivo stage allowed us to reach an important insight, which is that the gut acid actually neutralizes the phages almost to extinction,” says Elinav.

For the first time, we [were] able to develop a precision therapy that would target a group of commensals within this huge and divergent microbiome without impacting the surrounding critically important microbial ecosystem.

—Eran Elinav, Weizmann Institute of Science and the German Cancer Research Center

To compensate, they used antacids as a buffer to protect the phages. Their subsequent Phase 1 trial tested two five-phage cocktails against a placebo in 18 healthy humans who all took the antacid esomeprazole. The results showed the treatment was well-tolerated and safe, and that the phages persisted throughout the digestive tract. Analyzing the participants’ stool samples showed that the phages were still viable and at concentrations of about 1,000 times the levels expected to suppress K. pneumonia, says Elinav. Furthermore, he adds, the treatment did not harm other organisms within the microbiome.

“There’s a lot of work that needs to be done in terms of validating their findings in clinical cohorts and showing that this works with Crohn’s patients, but as a proof of principle, demonstrating that they can get these phages past the gut, I think is really encouraging,” says Bollyky.

Paul Turner, an evolutionary microbiologist at Yale School of Medicine who was not involved in the work, says the researchers used a good approach to identify problematic bacteria and develop phages to precisely attack them. “It’s going after a tough target, and it’s showing some compelling evidence that phage approaches may work,” he says, adding that this is a “holistic way to look at the problem.”

However, Turner says that the researchers’ iterative approach of using phages to address resistance problems after they arose assumes that a similar evolution would occur in the future, which is difficult to predict. “That’s a hard thing to completely convince an evolutionary biologist of because it’s kind of claiming that we’ve reigned in the evolution of these bacteria so that we don’t have to be worried about it.” He suggests another approach could be to try to select phages to attack a wider genomic range of K. pneumonia and to design them to attack binding sites associated with virulence or resistance mechanisms to “allow for the inevitability of the evolution of phage resistance,” instead of trying to prevent it.

See “Viral Soldiers

Elinav says that’s already on his list: He plans to investigate phages that attack a broader repertoire of K. pneumonia, and potentially the entire species, which would potentially yield IBD treatments for a larger population of people. Furthermore, he says that even as certain phages may be deemed clinically effective, researchers should continuously explore ways to improve their effectiveness.

Naama Geva-Zatorsky, a microbiome researcher at Rappaport Technion Integrated Cancer Center in Israel who was not involved with this study, writes in an email to The Scientist that more rigorous testing in humans is needed as a next step. “This is a beautiful research, novel, and first of a kind which thoroughly combined computational analyses and experimental validations in both mice and humans,” she says, adding that “more studies with similar meticulous assessments are needed.”

A Phase 2 clinical trial to evaluate the effectiveness of using a phage mixture to treat IBD in humans has been approved but not yet started, says Elinav, who cofounded a company seeking to commercialize phage treatments. The study authors conclude that in addition to the prospect of leading to a generalized therapy, their results indicate the potential to analyze the microbiomes of individuals and tailor specific phage cocktails to treat their illnesses.

Bollyky and Turner both agree that this work suggests broad and specific therapies may be feasible, but point out that targeting treatments to an individuals’ microbiomes would likely be inefficient. “Both approaches are potentially doable, but it becomes the investment and doing the latter requires a lot more infrastructure,” says Bollyky. In addition, Bollyky and Turner bring up the need to investigate other bacteria species associated with IBD.

While Elinav says that his group is interested in exploring phages to target other bacterial species, he’s cautious about investigating cocktails to target multiple bacteria simultaneously because of the complexity involved. “I wouldn’t start with targeting more than one bacteria at a time, but that’s my conservative two cents,” he says.

Describing this work as proof of concept, Elinav says he hopes that eventually an approach like that laid out in the new paper could be used to identify bacteria associated with a wide range of diseases—such as obesity, type 2 diabetes, cancer, and even neurodegeneration—in order to develop phage therapies to target them.