Autoimmunity, which occurs when the immune system attacks the body's own cells, is responsible for many human diseases, including type 1 diabetes. What triggers the autoimmune response that causes diabetes is not known. By building on the ideas of a former colleague, Alexander Chervonsky has made a specific connection between gut bacteria and the immune systems of diabetic mice.

As a postdoc, Chervonsky, now at the University of Chicago in Illinois, trained with the late Charlie Janeway, a champion of the idea that the immune system's tailored responses to specific invaders must be preceded by general — or innate — responses to 'non-self' invaders. Chervonsky wondered whether a similar general response might precede the specific attack on pancreatic cells that causes type 1 diabetes.

A strain of mice dubbed NOD — non-obese diabetic — is genetically predisposed to develop diabetes similar to human type 1 diabetes. To test Chervonsky's idea, his team engineered NOD mice that were missing the gene for MyD88, a protein with an important role in transmitting innate immune signals. These mice were protected against developing diabetes. But how?

Since the early 1990s, researchers have recognized that the microbes present in the animals' environment can influence whether or not NOD mice become diabetic. Chervonsky wondered what might happen if the mice lacking Myd88 were completely free of microbes — including the 'friendly' gut bacteria that colonize animals soon after birth and aid digestion.

Germ-free mouse pups were born and raised in a sterile environment with other germ-free mice. Germ-free mice lacking Myd88 exhibited a high incidence of diabetes just as normal NOD mice do. “I thought, great, that probably means that gut bacteria can be protective against diabetes,” says Chervonsky. But he needed more evidence. When the team added back a cocktail of a few known gut microorganisms to the germ-free animals, the incidence of diabetes decreased. The same thing happened when germ-free NOD mice were 'transplanted' with gut bacteria from NOD mice lacking Myd88.

Finally, led by co-author Jeffrey Gordon at Washington University School of Medicine in St Louis, Missouri, the team showed that there were specific differences in the types and amounts of gut bacteria present in normal NOD mice compared with NOD mice lacking Myd88.

Together, the results indicate that specific types of gut bacteria are protective against diabetes in NOD mice (see page 1109). “We think that, normally, the innate immune system controls the amount of 'friendly' bacteria living in the gut to keep it in balance,” says Chervonsky. “So the signalling that goes through MyD88 is controlling the proliferation of certain types of microorganism.” When MyD88 signalling is lost, the microbes that begin thriving somehow block the autoimmune signals that trigger diabetes. If those microbes are lost altogether (as in the germ-free mice), autoimmunity returns.

Exactly how the microbes protect the mouse against autoimmunity remains a mystery, but Chervonsky proposes a possible scenario. The pancreatic lymph node in which the autoimmune cells are activated is also the site of drainage for gut microbes. “It is likely that the microbes are trying to protect themselves and want to block immune responses for their own survival,” explains Chervonsky.