The possibility of reprogramming adult cells to behave like embryonic stem cells (ES cells) without overexpressing potential cancer-causing genes or relying on hard-to-control viruses has just got a bit closer. Publishing in Science, Shinya Yamanaka and colleagues from Kyoto University, in Japan, show that the reprogramming techniques they previously demonstrated on cultured mouse skin cells (see From skin cell to stem cell) also work on two other mouse cell types: liver cells and the epithelial cells lining the stomach1. Not only did these cells seem easier to reprogram, mice generated from the reprogrammed cells did not develop cancer.

Several labs have now reprogrammed fibroblasts, and because far fewer than 1 in 100 treated cells are successfully transformed, several stem-cell researchers had raised concerns that reprogramming does not work on fully differentiated cells but rather on rare stem cells residing undetected within the culture. That would make the reprogrammed cells less interesting scientifically and, potentially, therapeutically. To address this concern in his new study, Yamanaka used a genetic marking system that permanently labels liver cells once they differentiate enough to express albumin, and he found that these cells could be reprogrammed to so-called induced pluripotent stem cells (iPS cells) that can contribute to all cell types in an adult mouse. Yamanaka and colleagues also showed that the epithelial cells lining the stomach can generate iPS cells, and doing so requires a less rigorous screening system than that used with cultured skin cells, or fibroblasts.

“The old question in cloning was exactly the same: was Dolly derived from a fully differentiated cell?” says Rudolf Jaenisch of the Whitehead Institute, Cambridge, Massachusetts, who showed that mice can be cloned from terminally differentiated cells such as neurons. Yamanaka provides “good evidence” that reprogramming works in differentiated cells, says Jaenisch, but that conclusion assumes both a reliable labeling system and that only mature cells express the albumin gene. Yamanaka himself stops short of calling the initial cells fully differentiated: “Our data showed that lineage-committed albumin-producing cells can be reprogrammed,” he wrote in an email.

Tumours develop in about a third of mice created using iPS cells derived from fibroblasts treated with the four transforming genes originally identified by Yamanaka. However, no tumours were found in the mice created from iPS cells derived from stomach and liver cells. These mice were more likely to die in utero than those generated from fibroblasts, but the live-born mice appeared healthy. Compared with fibroblasts, viruses were less efficient at infecting the stomach-lining cells with the necessary genes, but the transformed cells contained fewer copies of the transgene compared with fibroblasts, perhaps because epithelial cells are more similar to ES cells than fibroblasts are.

Yamanaka also found that the transgenes do not need to be inserted into specific sites within the genome for liver and stomach cells to be reprogrammed. “This is encouraging to those of us who are seeking a non-viral means of generating iPS cells,” says George Daley of Children's Hospital, Boston, who has recently compared the efficiencies of reprogramming human fibroblasts from different sources2.