Less than a year ago, researchers showed that cultured skin cells could be transformed into a state almost indistinguishable from embryonic-like stem cells using a quartet of inserted genes. Amidst the excitement, basic-science researchers wondered whether the reprogrammed cells were more 'lucky' (that is, they happened to get enough copies of the four genes) or 'special' (that is, they arose from rare stem cells pre-existing within the cultured skin cells, or fibroblasts). Now, evidence from Rudolf Jaenisch and colleagues at the Whitehead Institute in Massachusetts sets the winner's cap on the lucky-cell hypothesis by demonstrating that terminally differentiated cells can indeed be reprogrammed1.

The paper, published this month in Cell, is not the first evidence for the lucky-cell hypothesis. Shinya Yamanaka, from Kyoto University in Japan, engineered cells to rearrange their genomes once they differentiated enough to make albumin and then showed that reprogrammed cells had the specific rearrangement2. But albumin-producing cells need not be fully differentiated, and that left room for doubt about whether a fully differentiated cell could adopt an embryonic-like state. To really settle the question, says Jaenisch, reprogramming must be accomplished in terminally differentiated cells in which the genetic rearrangement occurs without relying on an introduced gene.

As part of the process that generates a flexible immune system, B cells, the white blood cells that produce antibodies, naturally cut out a piece of their DNA in their final maturation step. The researchers thought they could use this quality as a marker to check whether the reprogrammed cells came from fully differentiated or immature cells. They found that both mature and immature B cells could be reprogrammed, and they verified that these so-called induced pluripotent cells were truly reprogrammed by using them to make chimeric mice.

Nonetheless, advocates of the special-cell hypothesis will find some solace in the paper. Though immature B cells could be reprogrammed using the same four genes already used to reprogram fibroblasts, the mature B cells required something more: researchers had to interrupt the gene expression typical of B cells, either with an additional factor (CCAAT/enhancer binding protein alpha) or by knockdown of the B cell transcription factor Pax5. After trying some 20 factors to convert B cells to a reprogrammable state, the researchers chose one that had been used to convert B cells to macrophages. “That worked almost immediately,” recalls Jaenisch.

In addition to addressing the basic science question of lucky cells versus special ones, the work has at least two sets of practical applications. One is to create better ways of studying autoimmune diseases by using reprogrammed mice generated from white blood cells that attack myelin or insulin-producing cells to simulate multiple sclerosis or diabetes pathology.

The other application will be to ease some of the difficulties of reprogramming experiments. The viruses typically used to reprogram cells don't infect B cells very well, so the researchers created mice that had reprogramming-ready cells — cells that already carried several copies of each of the four genes. These genes could be turned on at will by exposing the cells to the small molecule doxycycline. Under this system, the reprogramming rate is about 1 in 30 cells, says Jaenisch, a surprisingly high rate. Determining reprogramming efficiency rates requires many assumptions, says Jaenisch, because there's no way to directly measure how many cells are infected with the viruses. He hopes to use the reprogramming-ready cells to compare efficiency rates in other cell types.