The Microbial Health Factor
Just one molecule can make the difference in mediating a healthy immune response. Surprisingly, it comes from bacteria.
rillions of commensal bacteria cover almost all environmentally exposed surfaces of our bodies at all times. But what are they doing? And why? If you want to understand the impact of commensal organisms on mammals, a good place to start is with mice that are devoid of all bacteria.
When I started working on this problem in 2002, so few people were still familiar with the germ-free mouse models that I had to persuade a retired research technician to help me set up sterile chambers and teach me the ways of “sanitary engineering.” Rather than the old steel and glass contraptions that he had used in his day (50 years ago), we were able to procure nicely modernized chambers with plastic bubbles that held up to four mouse cages. After my first few chamber contaminations, I began to understand why researchers rarely use germ-free animals.
Germ-free animals were conceived of almost a century ago, but were not successfully raised until 1945. James A. Reyniers’ group at the University of Notre Dame was the first to successfully raise and study germ-free animals. Perhaps reflecting a new fervor over hygiene, researchers concluded that wiping a mammal clean of microbes might actually be a good thing. The adult mice grew enormous bellies, stemming from digestive problems, but other than that, they seemed just as healthy and lived just as long as typical mice.
In those days, science’s relationship with bacteria was adversarial—the main purpose of a microbiologist was to study infectious disease. No one seemed too curious about what the seemingly passive commensal bacteria were doing. Indeed, 20 Nobel Prizes have been awarded for research on the immune response to harmful microbes, from tuberculosis to Helicobacter pylori, the causative agent of gastric ulcers. But in the grand scheme of things, bacterial infections are rare and opportunistic. Of the over 300,000 known bacterial species and possibly millions more, only about 170 are known to be pathogenic in mammals.
When I trained as a microbiologist around the year 2000, the focus was still on pathogenic bacteria. But I became intrigued by the potential benefits of good bacteria. After all, we’ve coevolved with symbiotic bacteria for millions of years. The hygiene hypothesis, proposed in 1989 by David Strachan1, correlated lower environmental exposure to microbes—as seen in developed countries—with higher rates of allergies. The idea made sense to me. Commensal bacteria help keep pathogenic bacteria at bay, and in the late 1990s new research was beginning to show that symbionts also contribute to the development of the intestinal architecture. If bacteria were so crucial to development, what else might they do? Could they actually make us healthier? Challenging though it was, I was convinced the best way to learn about the systemic effects of bacteria was to start with mice that lacked them entirely.
Upon finishing my PhD in 2002 from the University of California, Los Angeles, I went to Dennis Kasper’s lab at Harvard Medical School. He was working on a prevalent commensal bacterium, Bacteroides fragilis. His lab had worked for years on the capsular polysaccharides that cover B. fragilis like hairs on a kiwi fruit. These surface carbohydrates are chains of repeating sugar molecules that function to give the bacteria a mucous-like barrier on its surface. Kasper had discovered that two of these eight polysaccharides have a unique zwitterionic structure: the molecules have both positive and negative charges on each repeating unit. While many bacteria are covered in polysaccharides, only a handful of species exhibit zwitterionic polysaccharides.
When I arrived at Kasper’s lab, I wanted to learn more about these polysaccharides and their properties. I mutated the B. fragilis genes that are involved in the production of polysaccharides to express different combinations of the eight surface polysaccharides. I was able to delete seven of the eight polysaccharides by deleting the gene’s promoter region, but despite hundreds of attempts, I never generated a viable culture of bacteria that lacked all eight. It was clear that the bacteria needed this sugar coating for normal function, but I wondered whether the polysaccharides might also be important because they offered something that the mammalian host lacked. The lab had already shown that PSA—the most prevalent polysaccharide of the eight—stimulated T cells of the immune system in the test tube. They had tested B. fragilis’s other zwitterionic polysaccharide, PSB, but had found that its stimulatory effects paled in comparison to PSA. Whether PSA influenced the entire immune system of animals was a question that could only be asked in a controlled manner in germ-free mice.
The heyday of germ-free animal work was in the 1940s and 1950s. But recently, other researchers dusted off the old germ-free mouse models and found problems beyond their big bellies and digestive troubles. With new cellular and molecular tools, researchers demonstrated that these animals had serious problems with their immune systems: antibody deficiencies, higher susceptibility to infections, reduced number of Peyer’s patches and germinal centers (the locations of lymphocyte activity), less active intestinal macrophages and a reduced number and cytotoxicity of intestinal epithelial lymphocytes. In 1992, Lex Nagelkerken, at the TNO Institute of Ageing and Vascular Research in The Netherlands, had found that germ-free mice had a reduced number of CD4 T cells compared to conventionally colonized mice. CD4 T cells are critical for regulating the immune response, activating both cellular and antibody reactions. We decided to use CD4 T cell levels as a proxy marker for a healthy immune system in germ-free mice.
We colonized one group of germ-free mice with whole B. fragilis and another group with a strain of B. fragilis that lacked PSA but displayed the seven other polysaccharides. To my delight, wild-type B. fragilis restored CD4 T cell levels to those of animals with hundreds of bacteria. In mice that were colonized with the mutant bacteria lacking the zwitterionic PSA, CD4 T cell levels were no better than in germ-free mice.
This was an important result. Not only was a single strain of bacteria able to restore healthy levels of CD4 T cells, but we also identified the specific surface molecule that mediated these effects. I checked for an effect on other arms of the immune system: the CD8 T cells that can directly kill other cells, and the B cells, which produce antibodies. These cells appeared not to be affected. It looked like the bacteria with intact PSA were inducing CD4 T cells specifically.
When we looked at the histology of spleens, which, along with lymph nodes, serve as sites for the generation of immune responses, we saw that germ-free mice exposed to B. fragilis without PSA lacked the well-defined follicular structures that are a hallmark of healthy immune cell development. Mice colonized with wild-type B. fragilis contained follicles in abundance. It was the first evidence that bacteria might play a role in the development of organs other than the intestine.
To double-check the specific role of this molecule, I purified PSA from the surface of B. fragilis. When I fed the germ-free mice the polysaccharide, they developed conventional CD4 T cell levels—in the absence of all bacteria!
The next question was whether PSA was stimulating all CD4 T cells equally or if one of the two branches was activated preferentially. At the time, the “helper” CD4 T cells were divided into two classes depending on the cytokines they secreted: T-helper 1 (Th1) cells, which activate the cellular arm of the immune system, and T-helper 2 (Th2) cells, which activate the humoral or antibody-producing B-cells. The balance between Th1 and Th2 cells is important for the proper function of the immune system. When we investigated the Th profile of germ-free mice, we found that they had an abnormal balance of T-helper cytokines. Germ-free mice produce large quantities of interleukin-4 (IL-4)—a Th2 cytokine—and very little inteferon gamma (IFNγ)—aTh1 cytokine—compared to conventional mice. But germ-free mice colonized with B. fragilis restored IFNγ levels to normal and reduced Th2 cytokines. Purified PSA was able to restore Th1/Th2 balance to the entire organism. What seemed to be an intrinsic feature of a healthy immune system was in fact completely controlled by a single bacterial molecule.
hortly after our findings were published2 an epidemiological study that extended the hygiene hypothesis caught my eye. The new study3 suggested that nonallergic autoimmune diseases such as multiple sclerosis, type 1 diabetes, and Crohn’s disease, were also on the rise in westernized societies. It occurred to me that there might be a possible role for B. fragilis in a wide range of immunologic diseases. The immune system is supposed to recognize foreign pathogens (such as bacteria) and eliminate them, while steering clear of healthy human cells. But sometimes the immune system can’t tell the difference between self and nonself, resulting in autoimmunity. One characteristic of this class of diseases is an imbalance in Th1/Th2 ratios, resulting in the immune system attacking host tissue. So if B. fragilis could correct a Th1/Th2 imbalance, perhaps it could also improve autoimmune diseases.
My autoimmune disease of choice was inflammatory bowel disease (IBD), a category of autoimmune disease that includes ulcerative colitis and Crohn’s disease in humans. Patients present symptoms that include abdominal pain, diarrhea, and rectal bleeding, caused by immune cell attack on the small or large intestine. These diseases affect about 2 million people in the United States, and that number is rapidly increasing. For decades, researchers had been looking for the pathogenic strains of bacteria responsible for IBD, to no avail. But several recent studies had pointed to the role of commensal—not pathogenic—bacteria in triggering IBD. With the benefits of B. fragilis in mind, I wondered: what if it wasn’t the presence of certain commensal bacteria triggering IBD, but the absence of protective symbiotic strains?
To test his hypothesis I colonized wild-type mice with B. fragilis, then induced IBD by introducing Helicobacter hepaticus—a bacterial strain known to initiate IBD in this experimental model. B. fragilis protected mice from IBD, but mice colonized with B. fragilis lacking PSA were not protected. The B. fragilis appeared to halt the autoreactive immune cells and prevent intestinal damage.
In 2006, I took an assistant professor position at California Institute of Technology. By then, researchers had discovered a new category of CD4 T cell that acted as a critical mediator of autoimmune diseases. This class of cells soon became all the rage in immunology circles. Th17 cells produce IL-17, a potent inflammatory T-cell cytokine associated with every known autoimmune disease. It became clear that immune reactions could actually be skewed toward Th1 (cellular), Th2 (humoral), or Th17 (autoimmune) pathways. Several studies found that Th17 reactions were involved in initiating IBD in mouse models. Was PSA inducing Th1 cytokines to shift the balance away from too much Th17 (and Th2)? If true, it would support my earlier study showing that PSA initiated the production of Th1 cytokines that reduced the higher Th2 response in germ-free mice. But what was the mechanism, and did PSA also suppress proinflammatory Th17 cells? I showed that PSA was still causing a proliferation of CD4 T cells, like in the germ-free mice of my earlier research, but also that another cell type called T regulatory cells was activated, and that it dampened inflammation by producing the cytokine IL-10. This cytokine was enough to suppress the pro-inflammatory IL-17 and protect the intestines from immune attack (see graphic above). I began to appreciate that B. fragilis was not inducing discrete immune responses such as Th1 cells, but was “shaping” a coordinated and complex immune profile to promote intestinal health.4
A few labs have recently been able to sequence the microbiota of healthy humans and IBD patients and showed dramatically reduced numbers of Bacteroidetes bacteria in IBD patients compared to healthy subjects. Currently, IBD patients are prescribed anti-inflammatory medicine, but this suppresses the entire immune system and puts the patient at a high risk of other illnesses and infections. In theory, B. fragilis as a probiotic therapy, or even administration of PSA alone, would have a more localized anti-inflammatory reaction in humans while cultivating the features of a functional immune system.
B. fragilis is certainly not the only important commensal bacterium in the human gut—it is merely the first one to be discovered with an immunomodulatory molecule. The Human Microbiome Project, an undertaking funded by the National Institutes of Health (NIH) to sequence the microbiota from hundreds of humans, has challenged itself with determining the relative quantities of all bacteria present in the human gut. With a known baseline of the bacteria present in healthy individuals, it will be much easier to understand which bacteria might be missing in diseased patients. Hopefully, the Human Microbiome Project will lead to the discovery of other beneficial bacteria.
For decades, scientists have been able to colonize germ-free animals with single organisms to evaluate their contributions to health. By pairing modern immunologic tools with these models, we’re starting to uncover truly novel effects of cohabitation with bacteria on human health.
Sarkis K. Mazmanian is an Assistant Professor in the Division of Biology at the California Institute of Technology. Sara W. McBride conducts research in Mazmanian’s lab.