Creating bacteria that produce the protein has a number of advantages over using the protein itself as the treatment. "The bacteria can secrete just the right amount of the protein in response to conditions in the host," says March. That could ultimately "minimize the need for self-monitoring and allow the patient's own cells (or the cells of the commensal E. coli) to provide the appropriate amount of insulin when needed," says Cynthia Collins, a bioengineer at Rensselaer Polytechnic Institute, in Troy, NY, who was not involved in the research.

In addition, producing the protein where it's needed overcomes some of the problems with protein-based drugs, which can be expensive to make and often degrade during digestion. "Purifying the protein and then getting past the gut is very expensive," says March. "Probiotics are cheap--less than a dollar per dose." In underprivileged settings, they could be cultured in yogurt and distributed around a village.

The researchers haven't yet studied the animals' guts, so they don't know exactly how or where the diabetic mice are producing insulin. It's also not yet clear if the treatment, which is presumably triggering intestinal cells to produce insulin, has any harmful effects, such as an overproduction of the hormone or perhaps an inhibition of the normal function of the epithelial cells. "The mice seem to have normal blood glucose levels at this point, and their weight is normal," says March. "If they stopped eating, we would be concerned."

March's microbes are one of a number of new strains being developed to treat disease, including bacteria designed to fight cavities, produce vitamins and treat lactose intolerance. March's group is also engineering a strain of E. coli designed to prevent cholera. Cholera prevention "needs to be something cheap and easy and readily passed from village to village, so why not use something that can be mixed in with food and grown for free?" says March.

However, the work is still in its early stages; using living organisms as therapies is likely to present unique challenges. More research is needed to determine how long these bacteria can persist in the gut, as well as whether altering the gut flora has harmful effects, says MIT's Prather.

In addition, recent research shows that different people have different kinds of colonies of gut bacteria, and it's unclear how these variations might affect bacterial treatments. "This may be particularly challenging when it comes to determining the appropriate dose of the therapeutic microbe," says Collins at Rensselaer. "The size of the population of therapeutic bacterial and how long it persists will likely depend on the microbes in an individual's gut."