Induced pluripotent stem cells make headlines today because of their potential applications in cell therapy and drug screening, but in August 2006, when Thomas Zwaka first learned about a technique from Kyoto University's Shinya Yamanaka that might make embryonic stem cells without embryos1, he was interested in basic science. Zwaka, of Baylor College of Medicine in Houston, Texas, thought it would be a great way to study pluripotency, the state in which a cell can differentiate into any tissue. As embryonic stem cells are already pluripotent, the only way to manipulate their state was to take factors out of the cells, he explains. With Yamanaka's induced pluripotent stem cells, there was the opportunity to see what happens when factors that potentially contribute to reprogramming are added to cells.

But Zwaka was worried about getting started. Yamanaka's system still hadn't been independently replicated, and Zwaka was reluctant to introduce an unproven system into his laboratory. But he took a chance and assigned a summer intern to try the techniques. She didn't have much cell culture experience, recalls Zwaka, “and the amazing thing was that it worked immediately, the first experiment”. The colonies that the cells were forming had all the tell-tale signs of embryonic stem (ES) cells. “And I knew from there on that this would just change everything.”

iPS cell mysteries

Reprogrammed cells Credit: Jaenisch lab

Induced pluripotent stem (iPS) cells have been minutely studied ever since their invention, but their fundamentals are still mysterious. How similar are they to ES cells? Will specialized cells generated from them adequately represent disease? Perhaps the key to answering these practical questions is the two most fundamental mysteries of all: how is it possible to turn a differentiated cell into a pluripotent one, and what is the pluripotent state?

In comparison to ES cell lines, iPS cell lines are astonishingly easy to make. The research community had established various cell banks to characterize, maintain and distribute human ES cell lines, but heavy paperwork, tricky techniques and the difficulty of obtaining funding all prevented the development of more ES cell lines. In contrast, researchers have shown again and again that finding new ways to make iPS cells is much easier than understanding what happens inside cells that are becoming pluripotent. To reprogram human cells, scientists typically need to add a combination of transcription factors. The original 'Yamanaka factors' are cMyc, Klf4, Oct4 and Sox2 (refs 2,3,4), although subsequent work shows that cMyc can be left out, albeit with a precipitous drop in reprogramming efficiency5,6. Besides the Yamanaka factors7, human cells can also be reprogrammed with a combination of Lin28, Nanog, Oct4 and Sox2 (ref. 8).

All reprogramming techniques described as of April 17, 2009 work by adding genes for the reprogramming factors, thus producing many extra copies of the transcription factors inside the cells. Expressing the introduced genes for several days seems to be essential, until, somehow, the genes awaken the cell's own pluripotency machinery. The best-studied techniques start with fibroblasts, a type of skin cell that is easily cultured and frequently stored in tissue banks, and then use engineered retroviruses to insert genes for the Yamanaka factors into the cells. Between 1 in 1,000 and 1 in 10,000 cells are reprogrammed. If you use nonintegrating viruses9 or plasmids10, reprogramming still happens but rates plummet. Try another cell type, like the epidermal keratinocytes11, and reprogramming rates soar. Rates also soar with the addition of small molecules that affect the epigenetic markers designating genes as expressed or silent12. Indeed, the addition of such molecules allows the reprogramming of neural stem cells, which already express Sox2 and Klf4, to be accomplished by adding only Oct4 (ref 13).

“Conceptually it's easy. Technically, it's not.” Sheng Ding, Scripps

Although several techniques have been reported for adding genes without the permanent genetic insertion of retroviral genes, many researchers would like to see techniques that don't use any DNA whatsoever. “Conceptually it's easy,” says Sheng Ding of The Scripps Research Institute in La Jolla, California, “technically it's not.” For instance, introducing the transcription factor proteins themselves would require not only getting the proteins into the cell but also getting them into the nucleus and making sure they persist in appropriate concentrations, explains Ding. Small molecules have been found that boost reprogramming rates, but it is not obvious how they might actually take the place of or stimulate transcriptional activity. One could imagine hunting for small molecules that can activate transcription factors, Ding says, but in differentiated cells, the genes for these reprogramming factors are silenced — the necessary proteins are just not there. That isn't stopping Ding and others from hunting out more circuitous passages to pluripotency.

[Editor's note: As this article went to press, Ding published a technique to reprogram mouse fibroblasts. See link to reading list at the end of this article.]

Reprogramming requirements

Nonetheless, organisms have evolved to prevent their cells from reverting to pluripotency, explains Yamanaka. “The three essential factors — Oct4, Sox2, Klf4 — have to reach their targets in order to make iPS cells, but those targets should be protected by many, many different walls or protectors.” To make reprogramming more efficient, people will need to figure out ways to tear down these walls, a fact that probably explains why small molecules that interfere with the maintenance of certain epigenetic markers can boost reprogramming rates.

Still, finding the factors that block reprogramming won't be enough to reveal the reprogramming mechanism. The reprogramming factors literally work very closely together (often as a multiprotein complex) to change the state of the cell. They bind cooperatively throughout the genome, and their relative concentrations are likely to matter. And, to make things even more complicated, the roles of the transcription factors probably vary during different stages of the reprogramming process, perhaps even at different points in the cell cycle. Nanog is essential for pluripotency, for example, but it isn't an essential reprogramming factor because Oct4 is able, somehow, to activate Nanog. “I don't think we can describe the hierarchy in a single figure,” says Yamanaka. The process, he says, is simply too dynamic. And, of course, there are other factors that sustain pluripotency besides the triumvirate of Oct4, Sox2 and Nanog. Such new players could be very important, says Yamanaka. “They may be able to lower the walls, or they may be able to break the walls.”

Last year, researchers led by Alex Meissner at the Harvard Stem Cell Institute, in Cambridge, Massachusetts, announced that they had identified a stable, partially reprogrammed state in which genes associated with pluripotency as well as those associated with the differentiated fibroblast state were active14. These cells could be nudged into pluripotency by inhibiting DNA methylation.

Kathrin Plath at the University of California, Los Angeles School of Medicine, has also studied this state, focusing on where Nanog, Oct4 and Sox2 bind to chromosomal DNA in fully reprogrammed cells, in partially reprogrammed cells and in ES cells15. For cMyc, which is dispensable for reprogramming, the binding pattern was very similar across all three cell types. But Nanog, Oct4 and Sox2, which are believed to form the core regulatory circuit of pluripotency, bound the genome at much lower rates in the partially reprogrammed cells, even though the levels of the proteins in those cells were actually higher than in the other two types. Something, says Plath, must be keeping these proteins from binding the genome: either some component the factors need to bind to the genome is absent or the state of the DNA and its associated chromosomal proteins is blocking their binding.

But Plath worries that this particularly stable, partially reprogrammed state may not be a true intermediate on the path to reprogramming. If it is, she wants to know how cells move into and out of this state. “We don't have the intermediate states. We have the beginning state, the end state and this partially reprogrammed state,” she says. “We should study the very early stage of iPS cells,” agrees Yamanaka, “but that is technically demanding because we have only a few cells that are on the way to iPS cells. Most will die or become something else.”

One tool for doing this are a series of reprogramming-ready mice, all genetically engineered to have different combinations of inducible pluripotency genes. Fibroblasts made from such mice could tease out which genes are needed when16. Andras Nagy of Mount Sinai Hospital at Toronto University thinks he may have another solution. In March, he and colleagues reported a way to reprogram human cells using a transposon instead of retroviruses17,18. The transposon inserts genes that can be induced by a small molecule and that can be completely excised with another small molecule. Most media coverage (including Nature) focused on the fact that the process leaves behind genetically unmodified but still reprogrammed cells. Nagy, however, says the system might be even more useful for tracking what's happening as cells reprogram. Using the transposon to reprogram fibroblasts is is about as efficient as the viral techniques. As with other reprogramming techniques, the reprogrammed cells can be mixed into a mouse embryo and grown into a mouse from which new fibroblasts can be generated. However, the secondary fibroblasts containing the transposon reprogram both very efficiently and very homogenously, creating populations of cells that could be studied throughout the time reprogramming is going on. Nagy is working to organize a consortium of labs to characterize the reprogramming process on a daily basis, tracking changes in, for example, gene expression, microRNA profiling, and protein content. This could provide both a map of what's going on and perhaps identify earlier stages in the reprogramming process that might be more useful for cell screening and disease application. After all, he says, pluripotency isn't an asset if you only want a particular cell type.

Lessons from nature

iPS generation is totally fascinating. But everyone knows there are missing pieces. Renee Reijo Pera, Stanford

Another route to understanding how reprogramming happens may be to study the reprogramming that occurs naturally in fertilized mammalian eggs. Before fertilization, egg nuclei express egg genes and sperm nuclei express sperm genes, explains Renee Reijo Pera, who studies reprogramming in early human embryos at Stanford University in California. “And in three days you have to go from those programs through reprogramming to an embryo.” After fertilization, this happens almost every time. But if a differentiated nucleus is transferred into an enucleated egg19, reprogramming happens in around 1 in 100 cells, she says. In both cases, reprogramming happens within hours. In addition to the much lower efficiency of iPS cell reprogramming, that process takes weeks rather than days and seems to require several rounds of cell division. “iPS generation is totally fascinating,” says Pera, “but everyone knows there are missing pieces.”

“A lot of the things that we do to a fibroblast are very different from the things an oocyte does to a nucleus,” says Zwaka, but the steps that cells go through might have some parallels. For example, Pera explains, small molecules that boost the efficiency of iPS cell generation were originally used to boost the efficiency of making ES cells through somatic cell nuclear transfer, which requires the generation of an early embryo. Zwaka is studying how gene expression changes as cells in mouse embryos become ES cells. He's been able to do similar experiments with iPS cells and is finding similarities. “This is important because we can actually compare not just the end product, but also how we arrive there.”

That could help elucidate or define the final state. Researchers are not at all certain that iPS cells are ending up in the same 'ground state' as ES cells or even that they are ending up in the same state as each other. “When you reprogram a blood cell versus a fibroblast, what are the differences, what are the similarities? Is there memory of the donor tissue? Once you've gotten to the iPS state, if you've gotten there via retrovirus or adenovirus or a combination of viruses and chemicals, do you get to the same state or different states?” asks George Daley of Children's Hospital Boston and a past president of the International Society for Stem Cell Research. Daley has argued that the research community should collectively set standards to determine what characteristics iPS cells must display. It's not uncommon, he says, for iPS cells to appear completely reprogrammed and then fail more rigorous tests of pluripotency.

James Thomson of the University of Wisconsin–Madison believes that now that there are more techniques to make iPS cells, including his and others' techniques to reprogram human cells without leaving in the introduced genes16,17,18,19, researchers can start to tease out whether any variability in cells' behaviour or gene expression is due to tissue memory, the reprogramming technique, genetic differences or other factors. “Those kinds of questions are going to be answered in the next couple of years,” says Thomson.

It's already clear that not all cell types initially labelled as pluripotent are equivalent. Cells derived from amniotic fluid, testes and the outer layers of mouse embryos can all be induced, with more or less difficulty, to form a variety of cells representing a wide range of tissue even though they do not pass other tests of pluripotency. Even changing culture conditions can shift cells from one pluripotent state to another. Daley suspects that iPS cells could, in fact, represent multiple similar cell states. Unless that question is sorted out, laboratories will be less able to compare results, and determining the cells' potential practical benefits will be stymied. “I think it's hazardous for us to say that if something matches some criteria of iPS cells, then we'll call them iPS cells and not to really know what [we're] working with,” he says.

Related articles

Embryonic stem cells 2.0

Fertilized eggs reprogram adult-cell genomes

Further reading

See more induced pluripotent stem cell references at Nature Reports Stem Cells' blog, the Niche.