In contrast, once a mammal's cells have gone down the path of becoming bone or skin or brain cells, there is normally no turning back. They are said to be terminally differentiated. If humans could undo differentiation, though, doctors might not have to hunt for rare and elusive stem cells within the body or try to force stem cells from one tissue to regenerate tissue of another type. Instead an ordinary pancreas cell might be turned into a progenitor of the insulin-producing cells lost in Type 1 diabetes. Normal nerve cells could become a neurone factory for brain or spinal cord repair.
Investigations of this approach are just beginning, but early results are both encouraging and intriguing. Harvard Medical School's Mark Keating and his colleagues first showed in 2001 that dedifferentiation in mammals might be possible by regressing mouse muscle cells with an extract from regenerating newt limbs. They attributed the reversion to proteins in the extract having switched on one or more genes in the cells.
Last year a group from the Scripps Research Institute also reported dediffer-entiating mouse muscle and then turning the cells into bone or fat. They used a small-molecule chemical that they found by trial and error and have named reversine, but as yet they are not sure how it worked.
Others are studying the natural environments, or niches, that stem cells usually inhabit within the body to figure out which environmental cues may tell stem cells what to do and when to do it. Allan Spradling and Toshie Kai of the Carnegie Institution of Washington have used this kind of information to control fruit-fly stem cells that normally produce the female's eggs. By manipulating niche signals, they could make the stem cells differentiate, then dedifferentiate again.
These kinds of results fuel speculation that such environmental signals may be crucial to creating and maintaining the stemness of stem cells. As Dov Zipori of the Weizmann Institute of Science in Rehovot, Israel, put it in a recent review article, a stem cell may turn out to be not an entity so much as a state--one that any cell could enter under the right conditions.