 Millions of patients with type 1 diabetes are short on beta cells. Pancreas transplants can help restore these insulin-producing cells, but there aren't enough donors to go around. That has led scientists to investigate stem cells, and whether they can be coaxed into pumping out insulin. The task has been difficult. But by turning to a 3D culture system, biologists from Kyoto University discovered a three-step process that's surprisingly efficient. The Japanese group began by searching for the best way to transform embryonic stem cells, cells with the potential to become any cell in the body, into definitive endoderm cells, or pancreatic precursors. They tested eight growth factors and small molecules in various combinations and concentrations, looking for cells that responded by expressing two key genes typical of endoderm. The scientists identified a trio of compounds that together converted more than 90% of cells in 2D culture. With these endodermal cells in hand, the researchers then experimented with another set of chemicals to push them into being pancreatic progenitor cells. The most successful blend used five factors, working synergistically, to convert about three-quarters of cells. Finally, to morph the progenitors into beta cells that produce insulin, the team screened 20 different molecules, settling on four in combination. Interestingly, one factor, dexamethasone, did nothing on its own, but it proved to be the necessary fourth ingredient to boost insulin production. The only problem was that the cells didn't release insulin in response to glucose, what they need to do if they are to be useful in diabetes treatment. So the researchers tried the four factor mix again, but this time let the cells grow in three-dimensional culture, instead of along the 2D plate. This 3D culturing mimics the microenvironment cells experience in the body during development, and it did the trick. The technique even worked with other stem cell lines. The results suggest that differentiation isn't just about feeding cells the right signals, but also is about structure. The discovery offers the tantalizing prospect of using this method clinically to restore functional beta cells in people with diabetes, as well as a new tool to develop diabetes drugs.