Ever since Shinya Yamanaka and Kazutoshi Takahashi showed that mature cells could be brought back to the pluripotent state of early embryos, researchers have been trying to perfect the recipe that launches reprogramming. Now Rudolf Jaenisch and collaborators at the Whitehead Institute show that supplying the key ingredients in the right ratio makes a big difference for making mouse induced pluripotent stem (iPS) cells that behave like embryonic stem (ES) cells.

Both Jaenisch's lab and the lab of his former graduate student Konrad Hochedlinger had created highly controlled systems to produce iPS cells from a variety of tissues. This involved making “reprogramming mice”, transgenic mouse strains whose cells carry inducible versions of genes for all four of the standard reprogramming factors (Oct4, Sox2, Klf4 and c-Myc). Both labs made the two mouse strains by placing the same four factors together on a single genetic construct inserted into the same spot on the mouse genome. When cells from the mice were cultured, the reprogramming genes could be activated through the addition of the same small molecule.

Even though the systems were similar, the iPS cells produced from the two mouse strains were different. In the tetraploid complementation assay, pluripotent cells are injected into a special embryo that cannot develop normally and then implanted in a surrogate mother. Usually, the embryos die before birth. However, embryonic stem cells and very high-quality iPS cells occasionally give rise to live-born, breathing mouse pups. Although iPS cells containing the Hochedlinger construct passed many tests of pluripotency, they did not produce “all-iPS” mice in this assay. iPS cells containing the Jaenisch construct generated “all-iPS” mice at rates comparable to ES cells.

Given how alike the reprogramming systems were, the observed distinctions were surprising, says Jaenisch. They also offered a unique opportunity. “Because both systems were so well controlled, you could really compare apples with apples,” he says.

Though both constructs encoded the same genes, the genes were placed in a different order and joined by different sequences. Both constructs produced similar amounts of messenger RNA, but protein expression was very different. Compared to the construct designed by Hochedlinger's lab, the construct from Jaenisch's lab produced 5 to 15 times more Oct4 and Klf4 and about half as much Sox2 and c-Myc. When Jaenisch's team added extra vectors producing Oct4 and Klf4 to cells expressing the Hochedlinger construct, the resulting iPS cells could produce “all-iPS” mice.

In addition, Hochedlinger's lab had previously noted a striking epigenetic signature distinguishing their iPS cells from the embryonic stem (ES) cells used to generate their reprogramming mice. The Dlk1-Dio3 gene cluster, which is usually 'imprinted', reflecting paternally inherited epigenetic patterns, was aberrantly silenced in the iPS cells. This was not consistently the case with cells from the Jaenisch reprogramming mice; however, follow-up experiments showed that high expression of Oct4 and Klf4 along with low expression of Sox2 and c-Myc were more likely to produce iPS cells that maintained normal imprinting in the Dlk1-Dio3 locus. (In contrast to work from other scientists, Jaenisch's experiments indicated that silencing at this locus is not absolutely associated with reduced pluripotency.)

What is clear, says Jaenisch, is that the relative ratios of reprogramming factors are important and that the standard methods for creating iPS cells leave these ratios to chance. “The stoichiometry at the very beginning of reprogramming really affects the quality of the iPS cells,” says Jaenisch. “If you generate iPS cells by viral transduction, you can't control for that. We need to be aware of that.” In fact, says Jaenisch, findings such as these may indicate that at least some differences observed between iPS cells and ES cells are not inherent to the reprogramming process but are instead due to technical aspects of current reprogramming protocols.