Lab mice are notoriously clean: To prevent introducing anything that could disrupt results, researchers house them in sterile conditions, and the food and water they ingest are monitored for extreme quality.
This also means that the mice’s microbiomes — the cohort of microorganisms that reside within all animals and influence many body processes — have been controlled for. While that environment can be conducive to studying a host of conditions, the findings may not really be relevant for people who don’t live under such regimented conditions.
That’s the argument from researchers behind a new study published Thursday in Science. They describe a new mouse model they created that is genetically a lab mouse, but its microbiome is derived from a wild mouse.
The new model may be an important tool as researchers continue to grapple with how best to make meaningful discoveries in mice that will translate well in humans. Much more often than not, drugs that show tremendous promise in mouse studies are taken into clinical trials only to have the products fail. Billions of dollars and several years are lost to such unsuccessful efforts.
This new study shows that “some of the differences weren’t the mouse’s fault, but the scientists’ fault for how we were housing them,” said David Masopust, an immunologist at the University of Minnesota, who wasn’t involved in the study.
Masopust and other scientists have created such models before, including by placing lab mice in the same cage as pet shop mice. But the idea has been slow to catch on.
“It’s a different way of doing science and not every mouse is standardized [in the same way],” Masopust said. “I look at it as a strength and not a curse because humans are not standardized.”
The researchers behind the new work took a different approach. They transferred embryos of a common breed of lab mouse into the wombs of wild mice, in a kind of surrogacy, to make sure that the entire microbiome was transferred over.
“The early effect of the microbiome may already happen in utero, and then during birth a lot of microbes are acquired and then even more during early childhood,” said Dr. Barbara Rehermann, an immunologist at the National Institute for Diabetes and Digestive and Kidney Diseases and the senior author of the study.
They repeated this process several times and let the new generations of mice also procreate to create a colony of what they called “wildling mice.”
The mice retained the genetic makeup of their lab mouse parents, including their jet black appearance, but had the microbial makeup of their surrogate mothers. This included not only bacteria, but also fungi and viruses. And in a departure from other mouse models, which are kept disease-free, some of the microbes in the new model were also disease-causing germs.
Now came the real test: “Are these mice better at predicting clinical responses?” asked Dr. Stephan Rosshart, an immunologist who had been in Rehermann’s lab at the NIDDK, but is now launching his own lab in Freiburg, Germany.
In order to find out, the scientists decided to give the new mice two treatments that had shown preclinical promise but that had ultimately failed in human trials.
One such treatment was an antibody that had been developed for autoimmune diseases such as multiple sclerosis. Preclinical work had shown the antibody could help increase the population of a type of cell that regulates the immune system. But a Phase 1 clinical trial went horribly wrong: The treatment triggered an uncontrollable immune response unlike the one the scientists hoped for, and two of the trial participants sustained life-threatening side effects.
The other treatment was one to prevent septic shock, a sometimes fatal side effect of infections. Mouse studies for sepsis treatments have rarely produced good results in humans, according to Rehermann and Rosshart, who wanted to see if their new model could do better. In fact, in a human trial of the treatment, the drug actually seemed to increase mortality, and several people in the trial died.
The researchers compared the treatments in wildling mice and traditional lab mice. When it came to testing the antibody treatment, Rosshart remembered sitting in front of the machine that would tell him whether the treatment increased the immune cells. He was hoping it wouldn’t, if the mouse were truly predictive. “I was so nervous, I was really feeling sick,” Rosshart recalled.
Samples from the wildling mice immediately predicted the desired immune cells were not going to increase in number, whereas previous work had suggested they would.
The same was true when Rehermann and Rosshart tested the sepsis treatment. Lab mice showed a really high survival rate, with nearly 80% alive four days past treatment. But the wildling mice told a different story: Only half the mice survived on day four, and by day five, only 40% survived.
“Our mice, in this case, might have been able to predict people’s response,” Rosshart said.
In a related perspective piece also published Thursday, Weizmann Institute of Science microbiome researchers Dr. Eran Elinav and Samuel Philip Nobs wrote that using these mice in preclinical research could “reduce the number of failed clinical trials, prevent unnecessary risks for patients, reduce unnecessary animal use, and preserve research resources.”
However, there could be added lab maintenance costs. The wildling mice would have to either be housed in a completely separate facility or researchers would have to take extra precaution to keep these mice quarantined from their “clean” lab counterparts.
“It definitely raises logistical issues and I think people understand with scientific evidence that this [model] really helps with the science,” said Dr. Lynn Bry, an infectious disease researcher at Harvard Medical School, who was not involved in the study. “But I think good programs will understand that this is a tool that can really help advance the science and can help advance clinical applications.”
For those in the microbiome field, the results of the study seem to underscore the importance of the bugs that inhabit our bodies.
“I think what this work has demonstrated that when we look at 99% of clinical trials that fail to translate into human impact, at least some of the reason may lie in the microbiome,” Elinav said.
Ultimately, Rehermann and Rosshart don’t see their model overtaking the ones that already exist. “We don’t want to create the impression that we’re trying to replace [preexisting] models,” Rosshart said. “There’s a place for all of these models depending on your research question.”