Missing Microbes: How the Overuse of Antibiotics Is Fueling Our Modern Plagues (22 page)

To explore how the practice of subtherapeutic dosing with antibiotics affects development, we began a series of laboratory experiments on mice. This has been the most exciting work of my career.

13.

… AND FATTER

Why do antibiotics make animals bigger and fatter? The goal in our study was to re-create the weight and size increases observed in farm animals in our lab and then tease out the principles for why they occur. It took a big team to address these questions, but several scientists played key roles: Ilseung Cho, a physician and a fellow in gastroenterology; Laurie Cox, a graduate student whose dissertation project revolved around the mouse models and who at age fourteen had started working with bacteria for her father’s company, which made products for clinical bacteriology labs; and postcollege student Yael Nobel. Without such intelligent and dedicated trainees, I could not have tested any of my ideas. And there were many others who joined in the quest, from high school and college students working during the summer to college students doing independent research and visiting scholars from around the world.

In 2007, after a number of attempts to get the model going, we began our first complete set of experiments on farm practices by adding four different subtherapeutic antibiotic treatments, which we called STAT, to the water bottles of mice. We only looked at females because they don’t fight as much as males, making work easier for us. The early results were not promising; there was no weight difference between the STAT and control mice.

When Ilseung’s research committee was told that the mice were not gaining weight, one of our experts asked, “What’s happening to their body composition?” He was referring to the proportions of fat, muscle, and bone. We didn’t know.

“Why don’t you DEXA them and find out?” he asked.

DEXA them? The term refers to dual-energy X-ray absorptiometry, a test given to women to determine their bone mass and risk for osteoporosis. But DEXA also tells us how much fat is in the body and how much muscle.

This suggestion turned out to be critical. We discovered that all four groups of STAT mice had about 15 percent more fat than the controls, differences that could not be explained by chance alone.

We had our first evidence that antibiotics were changing metabolism, affecting body composition. The STAT mice were making more fat and had about the same amount of lean muscle as the control. We also had an unexpected finding: at seven weeks of age, three weeks following the start of the antibiotics, the mice were putting on bone at an accelerated pace. More bone formation implies that they would become bigger, longer, and taller. But by ten weeks, all the mice had similar bone mass. The effect on bone showed up early only in those that were given antibiotics. In later experiments that I describe below, we also found bone effects, some of them lifelong. Again, this was not specific to a single antibiotic. If it was, one might think of it as a side effect of that one drug. But it was present across all the antibiotics tested. This work supports the idea that, in addition to better nutrition and clean water, antibiotics may be part of the explanation of why people are taller than ever.

We now had evidence that STAT changes early development but still did not understand how it happened. How did adding antibiotics to the water cause these developmental effects? What made the animals fatter and built up their bones earlier? We suspected that the drugs changed the composition of the intestinal resident microbes, so that is where we looked first by examining mouse poop. Fecal pellets represent the end product of everything that happens in the intestine and could be collected every day from each mouse. The pellets gave us a standard material to compare in the same mouse over time, and across the mice that were exposed to different antibiotics or not, and varied diets.

We also studied material from an upper region of the colon called the cecum after sacrificing the animals. Cecal content was important to our study because it showed us which microbes were present and active in the body, not only after elimination in the feces. Because it had to be removed surgically, we could collect it only once, when the mice were killed. Most intestinal content of mice and humans, whether in the colon or in the poop, is undigested dietary fibers, water, and bacteria; the DNA present is nearly all bacterial. We performed what’s called a universal bacterial 16S ribosomal RNA assay to learn more.

All bacteria share a gene that encodes 16S rRNA, which they need to make proteins. Although all bacteria have 16S rRNA genes, the exact DNA sequence substantially differs among bacterial species. The form in
E. coli
differs from that in
Staph
. So, by using the universal technique, followed by sequencing of the DNA products, we can take a census of “who is there.” It is similar to taking a census in New York or Chicago and asking how many teachers, lawyers, police officers, and schoolchildren live there. In this case, we are asking how many clostridia, bacteroides, streptococci, and so on are present, down to thousands of individual bacterial species. Based on the results of our census, we were able to address a number of important questions.

First, does the STAT treatment alter bacterial diversity? In other words, are the resident microbes of the antibiotic-treated mice as diverse as those of the control mice? Although both samples might be expected to have a lot of schoolteachers, students, and police officers, because they are common, will they have actuaries and piano tuners (rare professions) or had those dropped out?

We found that STAT, possibly because it is low dose, had no obvious effects on bacterial diversity. The same number of “professions” were present in the STAT-exposed and control specimens.

But what happens to the composition—the relative proportions of teachers, police officers, and so on—with STAT? We can take a census of who is there. For example, we would expect that the distribution of these professions in New York and Chicago would be closer in composition with each other than either would be with Delhi or Beijing. This is a model of what we find in the gut microbiome.

This is where things got interesting. STAT changed the composition of the intestinal microbial population, whether we examined the fecal pellets or the cecal contents. We expected that usual antibiotic exposures would change the mix, but we didn’t know whether very low doses of STAT would do the same. We found that they did.

But did they change the functions of the bacteria? The answer is yes. Most of the food you eat is digested and absorbed in your small intestine. Residual food that reaches your large intestine is mostly indigestible. But here your bacteria come to the rescue. Recall that certain microbes in the colon digest this material and produce what are called short-chain fatty acids (SCFA), which are absorbed in the colon. These SFCA represent 5–15 percent of the calories you take in every day. If your microbes were more efficient at extracting calories from this “indigestible” food, then you would be better nourished. You might get fatter.

We measured SFCA levels in cecal contents and found that they were significantly greater in the STAT mice than in the controls. That meant that STAT mice were getting more calories early in life from their microbes, just as their tissues were developing.

We next zeroed in on the liver, the body’s main metabolic factory. It transforms the food absorbed in the intestinal tract, including the SCFA, into useful products, including proteins, energy sources such as sugars and starches, and energy-storage molecules such as fat. We compared the genes expressed in the livers of STAT mice to those of the control mice.

We were right on target. The liver in STAT mice up-regulated the genes needed to make and transport more fat out to the periphery—the blubbery layers of fat animals. We knew that the STAT mice were putting down more fat and that it had to come from somewhere. The liver made sense. It is strategically interposed between the intestinal tract, where energy is acquired or generated, and the adipose tissue, where fat is stored.

*   *   *

Our next experiment, planned and carried out by Laurie, examined in more detail what happens when mice get antibiotics (we chose penicillin) very early in life. In Ilseung’s experiment, animals got the drugs when they were weaned, about twenty-four days after birth. This is equivalent to at least twelve months for a human baby. Now Laurie gave antibiotics to the mothers during their pregnancy, so that their microbes, including those in the vagina, were altered from the get-go. The infant mice began life exposed to an altered microbiome, and we continued to give them antibiotics. As we predicted, the mice exposed at birth grew more than those exposed at day twenty-four. That became our standard way to conduct experiments.

Next, Laurie conducted an experiment that zeroed in on
when
the mice begin to get fat. Mice grow rapidly from the time they are born. Would they be putting on the extra fat in their youth or did it take a while? The results of the experiment were clear. In males we found a difference from controls at sixteen weeks, and in females fat showed up at twenty weeks (middle age for a mouse). But in both sexes, once it was there, the increased fat persisted for their entire life span.

Subsequently Laurie looked at which species of bacteria were prevalent in these young mice. At four weeks, control animals were dominated by
Lactobacillus
, the bacteria originating in their mother’s vagina. This was expected because the animals had just finished nursing, a time when, in both mice and humans, lactobacilli dominate.

But in the STAT group most of the lactobacilli were gone, replaced by other groups of bacteria. Since the changes in body composition were detected after sixteen weeks and the resident microbes were different at four weeks, we had a critical observation: changes in the microbiome preceded the changes in body composition.

Some elegant work done by my longtime friend and colleague Jeff Gordon at Washington University in St. Louis adds insight to our findings. Jeff has been a giant in the field of microbiome science, building on his years of research on how the gastrointestinal tract develops and functions. Jeff’s group studied mice with a deletion in the gene responsible for the production of leptin, the “feed me” hormone that helps regulate appetite and helps the brain decide whether to store or use energy. Leptin-deficient mice, called
ob/ob
mice, become markedly obese. Jeff and his colleagues asked whether the resident microbes of the
ob/ob
mice differ from those of their normal littermates. The answer was yes. Each type of mouse had different microbial populations in its guts.

Then Jeff asked if the microbes performed different metabolic roles. He transferred intestinal contents from the obese
ob/ob
and the normal mice into germ-free mice. These mice have thinner intestinal walls with fewer cells and do not gain as much weight. But when they are conventionalized and get microbes back, how well do they grow? Jeff’s finding, which made news around the world, is that resident microbes taken from obese mice caused the recipient mice to put on fat at an accelerated rate compared to the mice that received microbes from the normal-weight mouse donor.

But here is something to consider: the mice in Jeff’s experiments had a genetic defect that made them obese to begin with. That was the cause of the obesity; the change in the microbial populations was secondary. Although Jeff’s team had beautifully characterized the consequences of obesity on the microbes and their functions, I did not think they were addressing the root cause of obesity. Moreover, germ-free mice, which provide an elegant system for testing specific hypotheses about immunity and metabolism, are completely artificial. Yet although there are no natural germ-free mice or humans, we still can learn much about the fundamental principles of host-microbial interactions.

My own view was that antibiotic-induced perturbations in resident microbes early in life in relatively normal individuals might be the primary events that change host metabolism. (It would be about two years until we had more definite proof.)

Next, we asked what happens if we combine STAT with a high-fat diet. As we all know, our children’s diets have gotten a lot richer in recent decades, whether from sugary drinks or from high-fat foods. They are taking in more calories on average than kids did one and two generations ago. We know that mice get fatter on a calorically rich diet, but would STAT increase or decrease the trend, or would it just be neutral?

Laurie called this experiment FatSTAT, and again the results were exciting. As we expected, mice on the high-fat diet got bigger than animals on normal chow. But adding antibiotics made a significant difference. We had mimicked the manner in which modern farmers are raising their livestock. Males on the combination (fat diet and antibiotic) were about 10 percent bigger still, having gained both muscle and fat. But the most striking differences were in the amount of body fat: with the combination, males had about 25 percent more, but females had an astounding 100 percent more. The females on the high-fat diet gained about 5 grams of fat, whereas those on the fat diet and antibiotic gained 10 grams. They doubled their body fat. That’s a lot, considering their total body weight was 20–30 grams.

Thus antibiotics had an effect, high-fat diet had an effect, but together they were more than additive; they were synergistic. For the female mice, the antibiotic exposure was the switch that converted more of those extra calories in the diet to fat, while the males grew more in terms of both muscle and fat. We do not yet know the reason for these sex differences. Nevertheless, the observations are consistent with the idea that the modern high-calorie diet alone is insufficient to explain the obesity epidemic and that antibiotics could be contributing.

We asked another simple question that Laurie’s thesis committee suggested to us. Up to this point, we were keeping the animals on STAT for their entire lives. Would a few weeks of antibiotic treatment be enough for the weight gains to persist? This was a question important for our children’s future. If the weight gain happens only after long-term treatment, then maybe this isn’t relevant to our kids. Very few get lifelong antibiotics. But if short-term exposures cause the problem, this may be a way to explain our current epidemic. Most children are getting relatively short exposures of antibiotics for their ear and respiratory infections, especially early in life.