Anyone who’s ever visited a research lab that studies mice knows how the animals stink. But the mice housed in rows of large plastic bubbles in Jeffrey Gordon’s lab at the Washington University School of Medicine smell surprisingly pleasant. They’ve spent their entire lives in a sterile, protected environment, inhaling purified air. Because of their meticulous upbringing, they harbor none of the microbes that normally give mice their distinctive acrid odor.

But living free of the bacteria that colonize most animals has also had a profound effect on the mice’s development. They have less fat than their microbe-ridden counterparts and have to eat 30 percent more food to maintain their weight. Their hearts are 20 percent smaller, and they have immature immune systems.

For the last decade, Gordon, a microbiologist and director of Washington University’s Center for Genome ­Sciences, has been trying to figure out precisely why. He and his students have spent that time investigating the complex microbial world inside both mice and humans, attempting to determine how bacteria exert their broad influence on our health. Each of us contains roughly 10 times as many microbial cells as human ones. And while some microbes make us sick, many play vital roles in our physiology. They give us the ability to digest foods whose nutrients would otherwise be lost to us, and they make essential vitamins and amino acids our bodies can’t.

And yet, because the vast majority of these microbes die when extracted from their native habitat, they have been impossible to study and have remained a mystery. “This is completely unexplored territory that is likely to have a large impact on our understanding of human health and disease,” says George Weinstock, codirector of the Human Genome Sequencing Center at the Baylor College of Medicine in Houston.

Researchers in the emerging field of metagenomics are beginning to map that unexplored territory. New ultrafast DNA-sequencing technologies allow scientists to study the genetic makeup of entire microbial communities, each of which may contain hundreds or thousands of different species. For the first time, microbiologists can compare genetic snapshots of all the microbes inhabiting people who differ by age, origin, and health status. By analyzing the functions of those microbes’ genes, they can figure out the main roles the organisms play in our bodies.

Ultimately, researchers hope to find out precisely how microörganisms lower or increase the risk of contracting certain diseases. Armed with that information, physicians might one day use an individual’s microbial profile to diagnose a disease, or manipulate the organisms in our gut to treat or prevent health problems. “There are a whole host of properties that turn out to be dependent on the presence of healthy indigenous microbiota,” says David Relman, a microbiologist at Stanford University. “As we recognize the fundamental importance of our microbial genome, it becomes increasingly important to understand the makeup of these communities and the roles they play.”

Failed Diets Explained
Jeffrey Gordon is tall and lean and wears the academic’s uniform of khakis and a button-down shirt. He doesn’t seem the type to love the newspaper cartoon Cathy, whose main character has spent most of her comic life complaining about diets and bathing-suit shopping–but he has a framed print of a strip from last January on his office wall. The strip was inspired by a landmark paper Gordon published in 2006, arguably the first major functional finding in human meta­genomics. Dismissing various excuses for the failure of her diets, Cathy finally settles on what has to be the most bizarre excuse yet: “overly efficient intestinal microbes.” Amazingly, it’s scientifically justified.

In 2004, Gordon’s team published a paper describing a genetic survey of the bacterial populations of fat and lean mice. Mice genetically engineered to be obese, Gordon found, harbored different populations of microbes than lean mice. Two major groups of bacteria, the ­Bacteroidetes and the Firmicutes, dominate both the human and the mouse gut; the obese mice had a lower percentage of ­Bacteroidetes and a higher percentage of Firmicutes. In 2006, the researchers published a follow-up study of 12 obese people showing that the same pattern held true in humans. The differences seemed to be related more to weight than to genetics; after losing weight for a year on either a low-carb or a low-fat diet, the obese subjects had gut-microbe profiles that more closely resembled their lean counterparts’.

In both studies, the researchers used a genetic surveying method known as DNA barcoding. First they created a genetic soup containing DNA fragments from all the microbial species found in a sample taken from one of their subjects–whether mouse or human. Then they sequenced a small segment of DNA that occurs in a slightly different version in every species. Analyzing these “bar codes” allows researchers to gauge the number of different types of bacteria in a sample, even if some of those bacteria have never been cultured or sequenced before.

The finding that the microbial populations of lean and obese people differ raised a tantalizing question: could gut bacteria affect a person’s weight? To answer that question, Gordon and his collaborators needed more than just information on the different types of bacteria in the gut; they also needed to figure out whether the populations of gut microbes in obese and lean mice actually functioned differently.

The researchers attempted to sequence as much of the bacterial DNA from each mouse as they could. (Because this type of study analyzes DNA from hundreds to thousands of species, it is much more labor intensive than a traditional sequencing study, which analyzes fragments of DNA purified from a single species and then combines them like pieces in a linear jigsaw puzzle.) An analysis of the sequenced DNA revealed that the microbes from the obese mice had a higher percentage of genes involved in breaking down otherwise indigestible complex plant sugars that are common in the human diet. That means that animals harboring these microbes can more effectively squeeze the calories out of food. Cathy the beleaguered cartoon heroine was right: even if obese people eat the same amount of food as skinny people, they may be destined to gain more weight.

What’s more, this trait appears to be transferable. When germ-free mice had microbes from obese mice transplanted into their guts, they gained more fat than those with microbes from lean animals. “We don’t know what chemical signals mandate this change [in the microbial population],” says Gordon. “But we do know there is a dynamic relationship between the amount of fat you have and these bugs.”

Gordon hopes to eventually answer much broader questions about humans’ microbial inhabitants. How does the makeup of these communities contribute to a person’s health? What is the origin of each person’s distinctive microbial menagerie? Are a person’s microbes determined mostly by her diet, or by where she lives, or by some other aspect of her lifestyle? How do our microbial populations change over time? And perhaps most important, can we tinker with an individual’s microbial profile to improve his or her health? Ultimately, says Gordon, “we’ll get a much more transcendent view of ourselves as a supraorganism, with traits encoded by our human genes and by those in the genomes of our microbial partners.”

Sequencing Microbial Complexity
One floor below Gordon’s lab is Washington University’s Genome Sequencing Center, one of the primary sites of the Human Genome Project. The center houses more than 130 “traditional” sequencing machines, capable of reading five to six billion letters of DNA every month. During a recent tour of the facility, Gordon whizzed past these genomic workhorses and down a hall to the room that houses the center’s newest acquisitions: two sequencing machines made by 454 Life Sciences of Branford, CT. The machines are among the world’s fastest gene sequencers, each reading an impressive 100 million DNA letters during a seven-hour run (see “Sequencing in a Flash,” May/June 2007).

The 454 sequencers are at the heart of Gordon’s next project. Researchers hope to learn how to manipulate the way microbes affect energy storage and metabolism–to predict and perhaps reduce the risk of obesity, or to aid people who are undernourished. To understand these effects, ­Gordon plans to compare the microbial profiles of family members–obese and lean siblings and their ­mothers–­in unprecedented depth. That’s possible thanks to the new machines, which can sequence hundreds of thousands of pieces of DNA in a single experiment; older machines can handle just a few hundred. Only after researchers have generated microbial profiles of many different people will they be able to gauge the normal variability of microbial profiles in people of different ages and origins. That, in turn, will help them determine which specific microbial changes can be linked to illnesses or other health issues.

A comprehensive effort to catalogue human microbial populations is far too large for a single lab to undertake. The National Institutes of Health acknowledged that in May by designating the human microbiome a “Roadmap initiative.” That means that significant funding will be available to support research in this area over the next five years. Scientists hope the initiative will ultimately blossom into a microbial version of the Human Genome Project. The project will be challenging. “Even though a microbial genome is one-­thousandth the size of the human genome,” says Baylor’s Weinstock, “the total number of microbial genes in [the human] body is much greater than [the number of] human genes, because you have so many different species.”

The success of the project will depend not just on ever-faster sequencing technologies but on new techniques for analyzing all that data. Scientists can examine a meta­genomic sequence in two ways: by studying microbial species separately and by studying the community of different species as a whole. The first approach involves piecing together individual genomes to deduce the roles that different species play in the gut. Most genomic-analysis tools, however, have trouble with the genetic soup that constitutes a meta­genomic DNA sample. The second approach is to look at the genes from an entire microbial community at once, without trying to analyze how they fit together into genomes. This approach gives a better picture of how bacterial communities may have evolved to function as a group, but it has its own limitations. Metagenomic studies of the ocean and other ecosystems have already revealed an unexpected bounty of genetic diversity; many of the genes uncovered are entirely novel and their functions entirely unknown. So scientists will also need better ways to predict these genes’ functions.

Taking Microbial Medicine
In an abandoned bubble cage in the room housing ­Gordon’s sterile mice sits a carton of yogurt. Yogurt is full of living bacteria that are meant to be good for you, and ­Gordon is testing “probiotic” theories about the beneficial use of microörganisms. Technicians feed yogurt to both sterile mice and those that have been purposely infected with one or a few species of bacteria, whose effects on their health and microbial profiles the researchers are trying to gauge.

“Maybe microbes themselves, or microbe-derived chemicals, could become part of our 21st-century medicine cabinet,” suggests Gordon. He envisions a day when a routine doctor’s visit would include an analysis of our microbial inhabitants. “When we eat, should we consider caloric and nutritional value of food as absolute, or should we consider it based on an individual’s microbial community?” he asks. Microbial profiles could also be used in diagnosing, and perhaps even treating, specific diseases.

Metagenomic studies could shed light on public-health issues, too. Microbiologists hypothesize that changes in the way we live–cleaner drinking water and growing use of antibiotics, for example–have changed which microbes colonize our bodies. This in turn may explain, at least in part, why many developed countries are seeing an abrupt rise in certain diseases, such as asthma, which is characterized by an excessive immune response in the lungs. In the absence of some types of bacteria, the immune system may develop or function abnormally. Martin Blaser, a microbiologist at New York University, says that almost everyone used to be infected with the gut microbe H. pylori, which has been linked to ulcers. Now only 10 percent of children in the United States carry the bacterium. Preliminary studies conducted by Blaser link its absence to asthma.

While scientists have been able to connect some disorders, such as ulcers, to the presence or absence of individual bacterial species, they have been unable to identify a single microbial culprit for many others. It’s possible that changes in many microbial populations must occur together to boost the risk of contracting a particular disease. “Maybe it’s the structure of the community that plays a role,” ­Gordon says.

Back in the bubble cages, mice are eating, sleeping, and nursing their young. By showing how microbial populations affect their animal hosts, the studies taking place in the mice’s sterile quarters could eventually have profound consequences for human health. But for now, they represent the first tentative, tantalizing forays into a mysterious microscopic world.

Emily Singer is the biotechnology and life sciences editor of Technology Review.