Decoding bacterial chatter may clarify our relationship to their vast communities.

Inside the human gut, trillions of microbes are networking. They’re passing notes of collusion to friends, handing out business-card-like identifiers to the immune system, and signaling threats to competitors. Some messages are game-changers, causing bacteria to shift their behavior, produce a new compound, or abandon the gut altogether. Other signals confirm that all is well.

In the late 1990s, scientists began to appreciate the extent and importance of the microbes that live in the human body—on the skin; in the mouth, stomach, and intestines; even inside ears and eyelids. Using modern genetic techniques, they began taking a census of the body’s microbes, collectively called the microbiome, and discovered that they varied based on diet, health, and environment.

“At the beginning, we in the field had this idea that we’d be able to create one single reference data set for a healthy microbiome,” says microbiologist David Relman of the Stanford University School of Medicine. “Now we’ve tempered our expectations as we’ve realized the complexity. There will be no one answer to what makes up a healthy microbiome.”

Correlations have emerged between unique patterns of bacteria in the gut and everything from inflammatory bowel disease and obesity to asthma and depression. But the associations, even when based on large populations and strong data, do little to explain the mechanism by which changes in these microbial communities can alter human health.

“I think many of us are becoming a little bit frustrated with having correlations alone,” says Akiko Iwasaki, an immunobiologist and newly named HHMI investigator at Yale University. “It ultimately doesn’t prove anything.”

Researchers like Iwasaki are focusing on how bacteria and viruses in the gut communicate, and how they interact with each other, with the physical barriers of the body, and with the human immune system. Understanding both the structure of the microbial social networks and the nature of the messages reverberating around the gut, they say, are key to making connections between the microbiome and its effect on the human body. “What we’re interested in now is the full spectrum of communication between the human host and the microbes we house,” explains HHMI Investigator Ruslan Medzhitov, Iwasaki’s husband and frequent collaborator at Yale.

In the process of decoding the microbes’ chatter, scientists have discovered a variety of new molecules—compounds that protect the gut’s lining from the microbiome, chemical signals produced by gut bacteria to communicate with each other, and unique clusters of immune molecules whose job is to interact with the bacteria. Slowly, they’re learning how to make sense of the language of the microbiome.

Page 2 of 4

Passing Notes

The very idea that bacteria communicate with each other—rather than acting only as individuals—was discovered in experiments on the marine bacterium Vibrio harveyi. When a small throng of V. harveyi sloshes around in a glass of nutrients on a lab bench, the slurry is unremarkable at first. Over time, though, the mixture begins to glow as the bacteria multiply. In the late 1990s, HHMI Investigator Bonnie Bassler discovered that V. harveyi produce a molecule, AI-2, that the bacteria use to sense each other’s presence. They luminesce only when they sense enough of the AI-2, indicating that they’ve reached a critical mass. Karina Xavier, who joined Bassler’s lab as a postdoctoral researcher, wondered whether this so-called quorum sensing worked only within that bacterial species, or whether messages were also passed between species.

Bonnie Bassler demonstrates bioluminescence in a culture of Vibrio harveyi.

So Xavier mixed V. harveyi and Escherichia coli in a test tube and measured the glow. She found that the V. harveyi produced only 18 percent of the light they typically emit. The E. coli was slurping up the AI-2, so the V. harveyi bacteria never sensed the concentration required to initiate luminescence.

Xavier and Bassler went on to show that almost every type of bacteria—not just V. harveyi and E. coli—produces and consumes AI-2 at different levels. The AI-2 acts as a general “here I am” flag for bacteria. Communities of bacteria use levels of AI-2 not only to decide when to fluoresce, but also to coordinate more aggressive actions, like the production of antibiotics that kill competing bacteria. So by consuming the AI-2, E. coli and other bacteria can counter these tactics using the same strategy to protect themselves from offensive attacks.

“Once I figured out that this molecule mediated changes when you mix bacteria, I wanted to increase the number of species in my experiments,” says Xavier, now an HHMI international early career scientist at Gulbenkian Science Institute in Portugal. But when she mixed three, four, or five species of bacteria, the culture failed—one species would kill off the rest and the community would never reach a state of equilibrium suitable for studying AI-2.

When Xavier launched her lab in Portugal, she turned to a stable, multispecies bacterial home she didn’t have to create from scratch: the gut. “It was exactly what I needed to study the communication among many species.”

In the gut, Xavier hypothesizes, AI-2 is just as important as it is in the marine environment of V. harveyi. She believes that it’s one of the signals that bacteria use to launch, or sync, the production of other self-serving molecules, such as disease-causing virulence factors, antibiotics that target competing bacteria, and substances that suppress the immune system. A species of bacteria can coordinate such actions once levels of AI-2 have reached a threshold. The higher the level of the molecule being made, the greater the likelihood of its having an impact—whether that means spurring a change to the immune system, infecting the wall of the gut, or killing off a neighbor.

Xavier has developed ways to increase or decrease the amount of AI-2 in the mouse gut so she can observe how changes to the signal alter the makeup of the microbiome. Her results are forthcoming.

“I really think that cell-to-cell communication between bacteria is incredibly important when it comes to the homeostasis of the gut microbiota,” she says. There are likely to be as many types of messenger molecules as there are bacterial species in the gut, she adds. Some, like AI-2, are passed between many species, while others are unique to one or a few.

Immune System Messaging

As bacteria mingling in the gut use signals like AI-2 to decide when to change their behavior, they’re also communicating with the immune system of their human host. Typically, when a bacterium or a virus sneaks into a host, the immune system recognizes a physical component of the microbe—its cell wall or tail-like flagellum, for example—as foreign. But this recognition isn’t sufficient to differentiate between pathogenic bacteria—those that cause disease and should be destroyed—and commensals—the harmless, or even beneficial, bacteria that should be left alone by the immune system. Particularly in the gut, pathogenic and commensal bacteria can appear outwardly identical. And a commensal bacterial species can even become pathogenic if conditions change. How does the immune system know?

Medzhitov likens the situation to a high-rise security system. “Detecting a person in a building does not necessarily mean they’re an intruder, since not all people are intruders,” Medzhitov says. “But if someone comes into the building through a window at night, then that might indicate the person is a burglar.”

Page 3 of 4

The immune system could, for example, recognize behaviors such as the obvious destruction of human cells in the intestines as being a sign of a pathogen. But it also could detect more subtle behaviors, like the production of bacterial metabolites.

“There are certain metabolites that are only produced by certain types of bacteria, or bacteria under certain conditions,” explains Medzhitov. By detecting these unique end products, the immune system can sense which bacteria are there.

The immune system has adapted to recognize these byproducts as signals to act. One such metabolic product, butyrate, has been linked with bowel and digestive health; low levels of butyrate are a sign of colon problems. Associations like this—between signaling molecules or metabolites and human health—are likely to be more accurate than associations between particular bacteria and health, Medzhitov believes.

So what does the immune system do when it recognizes a particular metabolite or combination of metabolites in the gut?

“Some immune responses are like a sniper, others are like a nuclear bomb,” Medzhitov says. One pattern of metabolites, he explains, could be specific enough that the immune system can systematically pick out one particular bacterial species to kill off. Other patterns, though, might suggest a completely out-of-whack gut microbiome and elicit a massive wiping clean of the bacterial communities so they can be re-colonized in a more balanced way. In both cases, the bacterial byproducts signal a class of molecules called Toll-like receptors (TLRs), which Medzhitov discovered in 1997. TLRs act as a middle man in the body’s alarm system against intruders.

Not all messages produced by the microbiome will lead to an immune response, however, a fact that is critical to the body’s ability to harbor the trillions of microbes. And Medzhitov has discovered that metabolites—and their interactions with TLRs—could be important to how commensal gut bacteria stay under the immune system’s radar. Metabolites produced by some gut bacteria suppress TLRs rather than activate them, ensuring that the bacteria are not recognized as pathogenic.

The body keeps the microbiome balanced by using physical barriers as well. They help provide a safe environment—free of immune system attacks—for commensal bacteria to thrive in the human digestive system.

“If you have a hundred trillion bacteria in your gut, how do you keep them where you want them?” HHMI Investigator Lora Hooper says. “That’s been a major question.”

In 2011, Hooper, at the University of Texas Southwestern Medical Center, discovered a microbe-free zone along the edge of the gut. A protein called MyD88, she found, helps enforce the territory, turning on components of the immune system—such as TLRs—if bacteria enter.

This June, Hooper published results from a collaboration with HHMI Investigator Beth Levine, also at UT Southwestern. Levine studies autophagy, which enables cells to self-destruct by degrading their internal parts. The researchers discovered that if invading pathogenic bacteria reach the cells lining the intestines, these infected epithelial cells have a unique autophagic pathway that kicks in and destroys them, thus preventing the microbes from spreading. Like the enforcement of the microbe-free zone, the pathway requires MyD88. When Hooper engineered mice that lacked MyD88, or proteins vital to the autophagy process itself, more bacteria traveled out of the gut to other sites in the body.

“We’re at a point now where we have lots of descriptive experiments confirming that the microbiome can affect the rest of the body,” says Hooper. “But we now need to translate that into cause and effect.”

By launching experiments focused on specific molecules, such as MyD88, that may play a role in this communication, Hooper aims to move beyond description and understand how molecular pathways function.

Inflammatory Remarks

While some researchers forge ahead with molecular biology approaches to decoding the cross-talk, others study the clinical implications of altering the microbiome. Their observations often lead right back to the biology of the microbiome–immune system interplay.

Immunologist Richard Flavell, an HHMI investigator at Yale School of Medicine, studies how multiprotein complexes called inflammasomes detect microbes when they invade the gut lining. In 2011, he and his colleagues reported in Cell that the absence of a particular inflammasome known as NLRP6 led to a profound change in the composition of the gut microbiota. Many microbes increased or decreased in abundance, and some began invading the gut lining. Those changes, in turn, led to increased susceptibility to inflammatory bowel disease (IBD). These same alterations, they reported in Nature in 2012, also caused increased susceptibility to obesity, type 2 diabetes, and fatty-liver disease—factors associated with metabolic syndrome, a condition that affects one-third or more of the population in the United States and other developed countries.

Flavell also knew that prolonged IBD is a major risk factor for colorectal cancer. Since NLRP6 plays a leading role in causing IBD when the microbiome is out of balance, he wondered whether there might be a link between the inflammasome and colorectal cancer. He decided to test that out in mice.

When mice with precancerous colon cells lacked NLRP6, the animals developed many more cancerous tumors than mice that had the protective inflammasome, according to a report by Flavell’s team in the May 21, 2013, Proceedings of the National Academy of Sciences. Like MyD88, the inflammasome protects the gut lining from bacterial invasion. If a bacterium can get through these defenses and colonize the gut lining, the immune cells in the gut are stimulated to release a molecule called IL-6. The scientists found that IL-6 acts directly on colorectal cells to cause cell division. In the case of precancerous cells, a sudden signal to divide can be the push toward cancer. The increased cancer burden was transmissible to genetically normal mice as an infectious disease—through an imbalance in the microbiome.

 “All these organisms interact with each other; they talk to each other,” Flavell says. “Once we can understand better how these interactions lead to human diseases, we can start trying to prevent those diseases.”

Page 4 of 4

Of course, bacteria aren’t the only microbes in that equation. At Yale, Iwasaki has discovered a link between multiple components of the inflammasome and how the body responds to the influenza virus. She’s shown that mice without the full set of inflammasome molecules can’t launch an effective fight against the flu. When Iwasaki went a step further and gave healthy mice a cocktail of antibiotics to wipe out their gut bacteria, these mice also were ineffective at fighting off the flu virus.

 “What the bacteria seem to be doing is turning on genes that induce the inflammasome response,” Iwasaki says. So having a gut full of commensal bacteria helps the immune system remain alert, not just to protect the body from pathogenic bacteria, but to detect invading viruses. Those TLRs discovered by Medzhitov are also vital to this link: without TLRs, Iwasaki’s lab group found, the inflammasome was no longer activated.

Crowd Connections

The challenge with studying the microbiome is that—as Xavier discovered before she turned to mice—it’s hard to keep bacterial cultures in the same balance seen in the body, where they interact with immune molecules and the lining of the gut. And mice don’t always mimic human gut conditions accurately. Pellets of mice chow don’t go down the same as a salad or a burger and fries.

"It’s like trying to study tiger behavior by watching one pacing around in a cage at the zoo instead of studying one in the wild,” says Rob Knight, an HHMI early career scientist at the University of Colorado Boulder.

Knight is a project leader for an initiative called American Gut, which aims to genetically sequence the microbes found in fecal samples from 10,000 adults around the world. By July 2013, the effort had raised $500,000 from more than 4,000 online donors, which Knight says is enough to start sequencing the microbiomes.

 "This is very exciting because we now get to see what microbes look like out there in the wild,” he says, "rather than in a few carefully defined cohorts."

In 2011, Knight published data showing that people could be classified, with 90 percent accuracy, as lean or obese based solely on their gut microbial populations. Of course, there are other ways of determining whether someone is obese, but such correlations—even if they don’t have immediate biological explanations—will lead to many early clinical utilities of the microbiome, Knight says. By testing particular aspects of the microbiome, clinicians may be able to predict who is at risk for a condition, how a disease will progress, or whether a particular diet or drug will be effective for someone.

Knight was part of a team of scientists who analyzed the microbiomes of 317 pairs of young twins in Malawi, a country in southeast Africa that is among the poorest and least developed in the world. The microbiomes, which differed even between twins, indicated which children were plagued by kwashiorkor, a form of protein-deficient malnutrition, according to their February 1, 2013, paper in Science. The microbiomes also predicted whose symptoms—including an enlarged liver and skin problems—would be reversed by a particular nutritional supplement.

“There’s a difference between what information can be useful and what information provides us with a complete understanding of the science,” says Knight. “And I think it’s important for the field to not worry about whether we have every bit of the microbiome system completely connected before we move toward applications."

Knight expects that tests to diagnose patients or shape their treatment plans will become available relatively soon. Creating personalized microbe-based treatments to change someone’s microbiome will take longer, he says.

To move toward personalized treatments, the field will need both correlative studies that link microbiome states to health—like Flavell’s research on metabolic syndrome and cancer and Knight’s on malnutrition—and mechanistic studies on how this connection plays out at a molecular level, Knight says. Each piece of research informs the rest of the field and helps shape new directions for researchers. “This is a time when the bench scientist has got to be talking to the clinician, who has got to be talking to the chemist, who has got to be talking to the ecologist,” says Relman. “If we just have a willy-nilly collection of lots of data, then understanding the microbiome will be slow going. But if we collaborate to design thoughtful experiments, we can get data that answers questions.”

  • 1
  • 2
  • 3
  • 4

Scientist Profile

Princeton University
Microbiology, Molecular Biology
Yale University
Cell Biology, Immunology
The University of Texas Southwestern Medical Center
Immunology, Microbiology
The University of Texas Southwestern Medical Center
Cancer Biology, Cell Biology
Yale University
Immunology, Molecular Biology
Early Career Scientist
University of Colorado Boulder
Show More