In type 1 diabetes, the body attacks and kills its own insulin-producing pancreatic beta cells, leaving patients dependent on insulin injections for life. Unlike the liver and skin, the pancreas does not have a well of adult stem cells at the ready to repair damage. When the supply of beta cells is exhausted, as happens in type 1 diabetes, there's no place to turn for more.
HHMI investigator Douglas A. Melton's sights are set on curing type 1 diabetes by regrowing beta cells, and he is trying to do so the only way he can, from embryonic stem cells. Guiding the ES cells through several developmental steps is a long, slow process.
“Figuring out how to tell these cells, which we know can do anything, what to do is a challenge,” he says. Researchers can tease stem cells into some types of neurons and blood cells, but Melton estimates that five to seven steps are required to create beta cells—and it has taken three years to figure out how to guide ES cells through just two of those steps.
This is stem cell research as developmental biology. Melton explains that the most immediate benefits of most stem cell research will likely come from what these cells can teach scientists about how cells differentiate, how tissues develop, and how disease occurs.
For example, once Melton discovers how to grow a beta cell, he could create stem cells from someone with type 1 diabetes and from someone else with healthy beta cells. Then he could grow both stem cells into beta cells and watch to see where the diabetic's cells go wrong.
By producing the disease in a Petri dish, scientists can run experiments—and even test drugs—in ways they never could in people. This is why the Harvard Stem Cell Institute has spent more that $6.5 million creating a lab to facilitate precisely that kind of work.
Thus, in the basement of the building where Melton conducts his research, a collection of boxy robots is at work. They hold plates barely larger than drink coasters, with each containing up to 384 cultures in separate tiny wells. Mechanical limbs move the plates around and drop different chemicals into each—creating, in effect, 384 distinct experiments. Next door, an automated microscope reader scans the results of these experiments; it takes up to two hours to scan all 384 wells on a plate.
That's still a lot faster than a postdoc. “In fact, a person couldn't really do that assay,” says Lee Rubin, director of translational medicine at the Harvard Stem Cell Institute. This is why most academic labs have traditionally studied only a few compounds at a time. Harvard researchers are using Rubin's robotic set-up—a stripped-down version of a drug-company screening lab—to ask a broad array of scientific questions. And it is how Melton figured out how to coax his embryonic stem cells two steps closer to a pancreas.