Despite the excitement surrounding stem cells' potential to perhaps cure disease or unlock the secrets of development, scientists know next to nothing about what genes, when activated, make a stem cell a stem cell. And it's not for lack of trying. The latest study on the topic may offer new clues, but it gives no definitive answers.

Mixed bag. Partial differentiation among stem cell populations such as these ES cells confound gene expression studies. CREDIT: UNIVERSITY OF WISCONSIN/MADISON

A year ago, two groups reported what they hoped would be a significant step forward (Science, 18 October 2002, pp. 597 and 601). Teams led by developmental geneticists Doug Melton of Harvard University and Ihor Lemischka of Princeton University used gene chips to compare the gene expression of embryonic stem (ES) cells, hematopoietic or blood-forming stem cells, and neural stem cells. Lemischka's group found 283 genes that were overexpressed in all three of their stem cell populations--presumably part of a genetic characterization of stemness. Melton's group found 230 genes that were overlapping in its populations. But there was a catch. The two sets of genes were almost completely different, sharing only six genes.

The new study only adds to the confusion. In a Technical Comment published online 17 October by Science, Bing Lim and his colleagues at the Genome Institute of Singapore describe similar experiments with ES cells, neural stem cells, and retinal stem cells. They found 385 genes that were overexpressed in all three cell types. However, only one of those genes is on both Melton's and Lemischka's lists.

So what's the problem? "There are a lot of caveats that need to go into interpreting these experiments," says developmental geneticist Leonard Zon of Harvard University. "The cell population you start with makes a huge difference in what is found in a microarray," he says, noting that isolating a pure population of stem cells is notoriously difficult.

Other technical problems may also confound the work, Lim notes. Key genes might vary their expression over time, he says, or perhaps the sought-after stemness genes are absent from the commercially available chips that all three teams used. In any case, he says, scientists have learned at least one thing: "The stemness gene is not a highly expressed one, if it exists," he says.

Both Lemischka and Lim say the results are worth following up. Both groups are working on functional studies of some of their candidate stemness genes, disabling them in populations of stem cells and observing how the cells behave. And even if the specific genes don't overlap, Lemischka says, it's possible that all three studies have identified common signaling pathways that are key to a stem cell's identity--an identity that remains frustratingly elusive.

--GRETCHEN VOGEL

Related sites
The Genome Institute of Singapore
A description of Zon's research
Lemischka's site
The Melton lab site