Researchers aim to harness microbes in our intestines to cure what ails us
By Bruce Goldman
They’re also little drug factories. They excrete myriad metabolic byproducts, which become a cocktail of chemicals coursing through our blood. Many of these have bioactive properties, said Michael Fischbach, PhD, Stanford associate professor of bioengineering and institute scholar at Stanford ChEM-H, and some are produced in sufficient amounts to produce a druglike effect, for better or for worse.
There’s growing evidence that the presence or absence of certain gut bacterial metabolites in a person’s body can cause, contribute to or even prevent a range of health problems.
“A single strain’s chemical output can add up to as much as 200 milligrams a day,” Fischbach said. “That’s about the same as the amount of active ingredient in an Advil tablet.”
Fischbach and other Stanford scientists from a variety of disciplines have shown that procedures as simple as adding or subtracting a single substance produced by a single gut bacterial strain can have a substantial medical impact. They’re developing tools to precisely define and manipulate our gut microbiota, or even individual genes within individual resident microbes.
And they’re working toward the creation of a defined “template” human gut microbiota that can serve as a scaffold for rebuilding our internal microbial communities to cure or ameliorate disorders such as kidney disease, inflammatory bowel disease, obesity and heart disease.
A single substance can make a big difference
Manipulating our gut microbiota is hardly a new idea. People have been eating yogurt for millennia. The whole point of ingesting probiotics is, after all, to upgrade our collection of gut microbes.
But that’s easier said than done. To begin with, we really don’t know which microbes would be ideal candidates for gut-microbiota membership. The varying composition of people’s gut microbiota, and the resulting difference in metabolic dynamics underway in different people’s guts, could complicate such simple designations.
But say we did know what’s good for whom. The same microbial lawn that protects against pathogenic invasions makes it tough for even the most desirable of microbes to take root. It’s never easy for a new kid moving into a new neighborhood already populated by battle-tested, streetwise residents who know their turf and know how to get exactly what they need, when they need it and who to get it from.
So it’s not surprising that probiotics are typically transient — within a day or two those freshly introduced bugs are already out of your system — or that their effect tends to be small and not highly predictable.
Fischbach and his colleagues have been taking steps toward what may prove in the long run to be a surer, more precise way to optimize a person’s gut microbiota. He and colleagues are on what he describes as a “hunt for microbiota-derived molecules of interest.”
This hunt can be exemplified by a study Fischbach conducted with Stanford colleagues Justin Sonnenburg, PhD, an associate professor of microbiology and immunology, Dodd Dylan, MD, PhD, an instructor in pathology, and several collaborators.
The scientists knew from previous studies that a gut-bacterial strain called Clostridium sporogenes is one of the few that can convert the chemical tryptophan, found in dietary protein, into a metabolite called indolepropionic acid, or IPA. Previous work had also suggested that IPA can strengthen the intestinal wall, preventing gut bacteria from getting into the bloodstream and possibly triggering a nasty inflammatory immune response — a characteristic feature of inflammatory bowel disease. IPA is also believed to be neuroprotective and has been considered a potential treatment for Alzheimer’s disease.
First, the investigators sought to determine the series of biochemical steps C. sporogenes takes to transform tryptophan into IPA, which was unknown. They used bioinformatics to nail down, for the first time, the genes coding for the enzymes involved in the tryptophan-to-IPA pathway. (Enzymes are protein machines that carry out virtually every biochemical transformation in a cell.)
“We’ve developed computational tools to sift through all the genes of the gut microbial community, and see just how a bug makes the molecules it’s making,” said Fischbach.
A bioinformatics application called ClusterFinder that Fischbach developed, since merged into an application called antiSMASH, relies on the premise that all of the enzymes that can perform a specific type of biochemical conversion tend to feature similar to identical biochemical structural elements — and that therefore the genes that encode these similarly functioning enzymes will also have similarities. Another key tenet of this kind of bioinformatics search engine: Genes that work together in bacteria are physically clustered together on the bacterial chromosome.
Fischbach and his colleagues trained an algorithm — effectively a digital bloodhound — to find them.
Finding the relevant genes in the IPA-production assembly line made it possible to create a mutant C. sporogenes strain in which one of those genes was disabled so that this strain could no longer produce IPA but was otherwise virtually identical to the IPA-producing strain.
Now the team was able to install one or the other bacterial version in living laboratories known as gnotobiotic, or germ-free, mice. These mice are raised in a sterile environment and have never been introduced to any microbial species that could colonize their guts. Researchers can therefore colonize the mice’s intestines with one or a few microbial species of interest to see what they do individually or how they interact in pairs or small groups.
The researchers found that germ-free mice into whose guts the normal C. sporogenes strain had been introduced carried copious amounts of IPA in their blood, while those harboring the mutant non-producers had negligible amounts of circulating IPA.
“We were able to show that by colonizing a germ-free mouse with the mutant or wild type, you can effectively toggle on and off that important chemical,” said Dodd. In addition, the mice with IPA-rich blood had lower levels of inflammatory immune cells, as well as less-permeable intestinal walls — a good thing, because it decreases the likelihood of a gut bacterium going AWOL — than those whose bloodstreams were devoid of IPA.
Increased intestinal permeability contributes to the symptoms of inflammatory bowel disease, a debilitating condition that affects an estimated 1.3 million adults in the United States.
“If we could find ways of increasing IPA levels in these people’s bodies, via some combination of seeding patients’ guts with C. sporogenes and ensuring adequate tryptophan intake, maybe we could decrease the severity of their symptoms,” Dodd said.
This is a drop in the bucket, he added. “It’s one example among hundreds of bioactive microbial-produced molecules that are medically significant.” While most of those remain to be characterized, several others besides IPA have been fingered, variously, as healthful or harmful to humans.
On the minus side, on the plus side
In a series of studies over the past several years, Stanley Hazen, MD, PhD, chief of preventive cardiology at the Cleveland Clinic, and his colleagues have implicated gut bacteria in the production of a substance called trimethylamine N-oxide, or TMAO, which is detrimental to cardiovascular health.
Searching for circulating chemicals whose levels in the blood are better predictors of heart disease than those now in use, such as cholesterol or C reactive protein, Hazen came up with TMAO. High levels of circulating TMAO, Hazen’s group has shown, predispose people to atherosclerosis, kidney failure, heart attacks, strokes and death, via an assortment of biological mechanisms.
TMAO’s production is known to require specific strains of gut bacteria. Those mystery microbes metabolize dietary choline and carnitine found in meat, eggs and fish to an intermediate substance, which the liver converts to TMAO.
Might it be possible to bioengineer, say, a TMAO-producing bacterial strain so it no longer produces the stuff, then introduce it to the gut in a way that allows it to outcompete the natural strain? That would allow us, as Russ Altman, MD, PhD, Stanford professor of bioengineering, of genetics, of medicine and of biomedical data science, has joked in reference to this goal, to “have our steak and eat it, too.”
Dodd said he, Fischbach and Sonnenburg are working on “figuring out what bugs make that troublesome intermediate and replacing them with doppelgangers that don’t make it.”
Generating the benign replacement strain or strains isn’t necessarily the entire solution, Fischbach added. There may also have to be some way of ensuring that whatever building blocks the “reformed” strain no longer metabolizes (and that could conceivably build up to toxic potencies themselves) get diverted to benign use instead. It may be necessary to introduce other bugs to slurp those up.
Back on the plus side, it’s known that various members of our gut microbiota can convert fiber — in essence, all the complex carbohydrates in our diet that we can’t digest, but that gut bacteria can — into substances called short-chain fatty acids that are a required energy source for cells lining our intestines and that, to boot, seem to exert a calming influence on our immune systems.
Stanford scientists led by Denise Monack, PhD, professor of microbiology and immunology, showed in a 2018 study in Cell Host & Microbe that propionate, a short-chain fatty acid produced as a metabolic byproduct by gut-resident members of the bacterial genus Bacteroides, protects lab mice against infection by Salmonella (whose people-infecting counterpart can cause typhoid fever or food poisoning) by diffusing into the pathogen’s cells and altering their acidity. The study authors suggest that boosting Bacteroides populations in the human gut may help control the spread of Salmonella and other pathogens.
“This is just the tip of the iceberg,” said KC Huang, PhD, associate professor of bioengineering and of microbiology and immunology, who was one of the study’s co-authors. “In the next few years, we may identify hundreds of such ‘therapeutic molecules’” produced by our gut bacteria, “and who produces them.”
In with the new
The intestinal lumen of a germ-free mouse is a great place to find out whether and how the manipulation of individual gut microbes, or specific genes in a microbe, can be tweaked to provide a medical benefit. But it raises a question: How does one stably introduce a single new species, however sculpted it may be to one’s nutritional or medical needs, into the immensely complex, fiercely competitive and — after millions of years of coevolution — incredibly clannish old-bug network holed up in the human lumen?
Paradoxically, it may turn out to be easier to just replace the whole gut ecosystem.
“We’ve learned some remarkable lessons from fecal transplants,” said Fischbach of the procedure in increasingly widespread medical use, in which the gut microbiota of a person with a health problem traceable to some fault in that ecosystem — say, infection by the deadly pathogen Clostridium difficile — is replaced by the microbiota of a healthy donor.
“The rate of adverse events is remarkably low,” Fischbach said. “I would have expected 1 in 100 or at least 1 in 1,000 recipients to have, for example, an immune reaction against the new bugs. But no! Not only that, the new microbiota often engrafts well. The recipient’s microbiota looks a lot like the donor’s did, even months after the transplant.”
This begets a counterintuitive, exciting conclusion, Fischbach said. “I would have thought it would have been easier to add or subtract one or two species to optimize a person’s gut ecosystem. But it might be simpler, from a stability standpoint, to replace the whole community.”
But fecal transplantation as currently practiced has its drawbacks. For starters, a potential donor’s gut ecosystem is an undefined combination of hundreds of microbial strains, some known and many of them unknown. So there’s no methodical way to optimize it to improve efficacy.
This problem plagues animal research, too. “You can show that a mouse’s outwardly observable characteristics change — for instance, it puts on weight, or sheds it — when you alter its gut-microbial contents via a fecal transplant. But now you’re stuck: Which specific bug in that transplanted microbiota was responsible? You have no idea. From that point it’s impossible to do anything more, because the fecal transplant isn’t defined.”
Then there’s the scale-up problem. It’s hard enough to find donors who have specific metabolic attributes you desire. But even if you’ve demonstrated that a particular human donor’s microbiota is adept at, say, reversing obesity or eliminating insulin resistance, you can’t treat an unlimited number of patients from one donor. And without knowing exactly what’s in there and how these component strains are interacting, you can’t just cook up huge quantities of it.
So Fischbach, Sonnenburg, Dodd, Huang and others are fashioning a work-around: a defined “model microbiota” to serve as a scaffold for purposed, customized designer microbial communities.
“Why not just completely replace somebody’s gut microbes with a community that’s built to spec?” Fischbach said.
You can generate such a defined “alpha-template” from the bottom up or from the top down, said Huang. “There are two complementary views on any engineering problem,” he said. “You can try to build a radio from components you understand. Or you can take the radio out of a car and try to figure out how it works.”
Fischbach and Sonnenburg are proceeding from the bottom up. “We’re assembling an entire gut community from scratch,” Fischbach said. “Something like 100 or 200 microbial strains are shared by pretty much every person’s gut. So we’re starting with them. We’re using them as a scaffold — putting those roughly 100 or so strains into germ-free mice, letting them get comfortable, then challenging that community by introducing a complete fecal sample from humans and seeing which new strains manage to get a toehold.
Every new member must be filling some not-fully-exploited niche in this ecosystem. We sequence the newcomers, find out who they are, add them to the community and do the same thing again, repeating until things get reasonably stable — the rate of strain gain and loss bottoms out.”
Huang is taking the top-down approach: suspending mouse fecal pellets in test-tube environments resembling that of the gut. He’s also seeing what changes occur when you systematically vary the culture medium (by, say, denying them a certain nutrient or by giving them more of it) or subtract one resident strain at a time. The idea is to generate a reduced, but stable, complex microbial community of 100 species or so whose metabolic characteristics are well understood.
Either way, once the researchers can derive a defined, stable, scalable scaffold, they can alter one bug at a time to build customized, synthetic, side-effect-free gut communities that predictably produce, or don’t produce, specific chemicals and can survive and thrive in a real human intestine.
“We don’t need to build custom communities for every person,” Fischbach said. “We could build one community for, say liver disease and another for inflammatory bowel disease or chronic kidney disease, analogous to different drugs that each treat these separate conditions in lots of people. One therapy can work in hundreds of thousands of people.”
But the one-step-beyond “abdominal medicine cabinet” is still many years away. The advent of a purposed, customized designer microbiota will have to await the patient collection of dozens or hundreds of “pill bottles” — the discovery and/or manipulation of individual gut microbes, or specific genes in a microbe, that can make a sustainable contribution to human health.
“We’re not very far along yet,” said Fischbach. “We’re making demonstrable progress, though. We’ve reached the jumping-off